technische universit¨at berlin - 3d.bk.tudelft.nl

Technische Universitat Berlin

Institute for Geodesy and Geoinformation Science

Faculty VI Planning Building Environment

Faculty VIStraße des 17. Juni 135

10623 Berlinhttp://www.igg.tu-berlin.de/en

Master Thesis

Estimation of Electric Energy Demand

using 3D City Models

Camilo Alexander Leon Sanchez

Matriculation Number: 336228

[email protected]

Supervised byProf. Dr.-Ing. Frank NeitzelDipl.-Ing. Thomas Becker

c© 2013

http://www.igg.tu-berlin.de/menue/institut_fuer_geodaesie_und_geoinformationstechnik/parameter/en/

mailto:[email protected]



Declaration

I, Camilo Alexander Leon Sanchez, matriculation number 3366228, hereby declarethat I have written independently my Master’s thesis on the subject

Estimation of Electric Energy Demand using 3D City Models

with no other help than the stated sources and aids which I have marked with propercitations as such.

Hiermit versichere ich, Camilo Alexander Leon Sanchez, Matrikulation Nummer3366228, dass ich meine Master-Arbeit auf das Thema

Estimation of Electric Energy Demand using 3D City Models

selbstandig verfasst und keine anderen als die angegebenen Quellen und Hilfsmittelbenutzt sowie Zitate kenntlich gemacht habe.

Berlin, August 12, 2013

Camilo Alexander Leon Sanchez

Camilo AlexanderLeon Sanchez

Beneficiario de Colfuturo 2010 i



Contents

Contents

List of Figures iv

List of Tables vi

Abstract viii

Zusammenfassung ix

Acknowledgements x

1. Introduction 1

1.1. Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2. Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2. Geographic Information Systems 3

2.1. City Geography Markup Language . . . . . . . . . . . . . . . . . . . 32.1.1. Thematic Model . . . . . . . . . . . . . . . . . . . . . . . . . 52.1.2. Extending CityGML . . . . . . . . . . . . . . . . . . . . . . . 8

2.2. 3D-Geo-Database for CityGML . . . . . . . . . . . . . . . . . . . . . 9

3. Energy 10

3.1. Methodologies used for Estimation of Electrical Energy Demand . . 113.1.1. Trend Method . . . . . . . . . . . . . . . . . . . . . . . . . . 113.1.2. Time-Series Method . . . . . . . . . . . . . . . . . . . . . . . 123.1.3. End-Use Method . . . . . . . . . . . . . . . . . . . . . . . . . 123.1.4. Econometric Method . . . . . . . . . . . . . . . . . . . . . . . 123.1.5. Advantages and Disadvantages of the Mentioned Methods . . 14

3.2. Occupancy Influence . . . . . . . . . . . . . . . . . . . . . . . . . . . 153.3. Electrical Energy Appliances . . . . . . . . . . . . . . . . . . . . . . 17

3.3.1. Food Preparation (ECFP) . . . . . . . . . . . . . . . . . . . . 183.3.2. Laundry (ECL) . . . . . . . . . . . . . . . . . . . . . . . . . . 183.3.3. Electrical Lighting (ECEL) . . . . . . . . . . . . . . . . . . . 183.3.4. Entertainment and Technology (ECEnT) . . . . . . . . . . . 193.3.5. Personal Computer and Home offices (ECPC) . . . . . . . . . 193.3.6. Miscellaneous (ECM) . . . . . . . . . . . . . . . . . . . . . . 203.3.7. Summary of Electric Model . . . . . . . . . . . . . . . . . . . 20

4. Comprehensive tools for Processing CityGML Files 21

4.1. citygml4j 2.0ea . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214.2. JTS Topology Suite 1.13 . . . . . . . . . . . . . . . . . . . . . . . . . 214.3. FME . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214.4. ArcGIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22


Beneficiario de Colfuturo 2010 ii



Contents

5. Estimation of Electrical Energy Demand using CityGML 23

5.1. Building Data Requirements . . . . . . . . . . . . . . . . . . . . . . . 235.1.1. Test Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245.1.2. Building Type . . . . . . . . . . . . . . . . . . . . . . . . . . 255.1.3. Estimation of Number of Storeys . . . . . . . . . . . . . . . . 265.1.4. Public Areas Inside a Building . . . . . . . . . . . . . . . . . 305.1.5. Dwellings / Inhabitants . . . . . . . . . . . . . . . . . . . . . 315.1.6. Workflow of the Estimation of Building Parameters . . . . . 33

5.2. Energy Appliances . . . . . . . . . . . . . . . . . . . . . . . . . . . . 345.3. Human Behaviour . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

5.3.1. Workflow of the Estimation of Energy Consumption . . . . . 36

6. Results 38

6.1. Building Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . 386.1.1. Storeys per Building . . . . . . . . . . . . . . . . . . . . . . . 396.1.2. Dwellings per Building . . . . . . . . . . . . . . . . . . . . . . 436.1.3. Number of Inhabitants . . . . . . . . . . . . . . . . . . . . . . 50

6.2. Electrical Energy Consumption . . . . . . . . . . . . . . . . . . . . . 526.2.1. Electrical Energy Consumption per Building . . . . . . . . . 536.2.2. Electrical Energy Consumption per Dwelling . . . . . . . . . 556.2.3. Electrical Energy Consumption per Inhabitant . . . . . . . . 56

6.3. Concrete Case of Study . . . . . . . . . . . . . . . . . . . . . . . . . 58

7. Summary and future work 65

Bibliography 67

A. Appendix A i

B. Appendix B iii

B.1. Survey Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii


Beneficiario de Colfuturo 2010 iii



List of Figures

List of Figures

2.1. The five levels of detail (LOD) defined by CityGML . . . . . . . . . 52.2. Example of buildings consisting of building parts . . . . . . . . . . . 62.3. Excerpt from the UML Diagram of the Thematic Building Model . . 72.4. Levels of Detail of the Building Model of CityGML . . . . . . . . . . 8

3.1. Energy Results for Cold Climates. Example of Occupancy Influence 16

5.1. Test Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255.2. Example of a LOD2 Building . . . . . . . . . . . . . . . . . . . . . . 275.3. Height dimension in Sections and Elevations . . . . . . . . . . . . . . 285.4. WMS Berlin buildings’ age . . . . . . . . . . . . . . . . . . . . . . . 285.5. Overlapping mismatch of WMS and buildings’ footprint . . . . . . . 295.6. Ground Surface feature of a building . . . . . . . . . . . . . . . . . . 315.7. Several designs of public stairs inside a building . . . . . . . . . . . . 315.8. Examples of dwellings . . . . . . . . . . . . . . . . . . . . . . . . . . 325.9. Activity Diagram for the estimation of Building Parameters . . . . . 335.10. Proposal of classification of Occupants behaviour . . . . . . . . . . . 355.11. Activity Diagram for the estimation of Energy Consumption . . . . . 37

6.1. Survey of the area of study . . . . . . . . . . . . . . . . . . . . . . . 386.2. Number of storeys per building classified by year of construction . . 396.3. Histograms of the estimation of number of storeys per building . . . 406.4. Example of buildings with and without without residential attics . . 426.5. Examples of additional roof constructions . . . . . . . . . . . . . . . 436.6. Histograms of the estimation of number of dwellings per building . . 446.7. Scatter plot of dwelling’s areas of surveyed buildings . . . . . . . . . 456.8. Scatter plot of dwelling’s areas of surveyed buildings according to its

year of Construction . . . . . . . . . . . . . . . . . . . . . . . . . . . 466.9. Histogram of the number of dwellings considering an average size per

year of construction . . . . . . . . . . . . . . . . . . . . . . . . . . . 476.10. Extract of the CityGML encoding standard for the declaration of

building parts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 486.11. Example of buildings with multiple uses . . . . . . . . . . . . . . . . 496.12. Example of buildings with multiple uses . . . . . . . . . . . . . . . . 496.13. Life style example for the electric lighting . . . . . . . . . . . . . . . 526.14. Extract of the activity diagram for estimation of energy consumption 536.15. Building’s energy consumption results (kW/H Year) for different life

styles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 546.16. Average dwelling’s energy consumption results (kW/H Year) for dif-

ferent life styles at the surveyed buildings . . . . . . . . . . . . . . . 566.17. Average energy consumption per person results (kW/H Year) for dif-

ferent life styles at the surveyed buildings . . . . . . . . . . . . . . . 57


Beneficiario de Colfuturo 2010 iv



List of Figures

6.18. Concrete example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 586.19. Scanned image of the poll done to a resident . . . . . . . . . . . . . 59

A.1. Run Configuration of Eclipse . . . . . . . . . . . . . . . . . . . . . . ii


Beneficiario de Colfuturo 2010 v



List of Tables

List of Tables

2.1. LOD 0-4 of CityGML with their proposed accuracy requirements . . 4

3.1. Energy flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103.2. Typical examples of single equation econometric models . . . . . . . 133.3. Advantages and disadvantages of the energy demand forecasting meth-

ods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

5.1. Semantic themes of the class AbstractBuilding . . . . . . . . . . . . 235.2. AbstractBuilding Class . . . . . . . . . . . . . . . . . . . . . . . . . . 245.3. Attributes available at the CityGML dataset of Berlin . . . . . . . . 245.4. Classification of residential buildings in the city model of Berlin ac-

cording to the ALK . . . . . . . . . . . . . . . . . . . . . . . . . . . 265.5. Height of Buildings based on the year of construction . . . . . . . . . 295.6. Classification of building age for the area of study . . . . . . . . . . 305.7. Relation between the number of inhabitants per dwelling place and

its corresponding area in square meters . . . . . . . . . . . . . . . . 325.8. Energy Appliances and their energy consumption . . . . . . . . . . . 345.9. Schedule ranges per energy appliance . . . . . . . . . . . . . . . . . . 36

6.1. Surveyed buildings classified by building’s year of construction . . . 396.2. Comparison of number of storeys per building between the observed

values and the estimated number of storeys in the implementation . 406.3. Comparison of number of dwellings per building between the observed

values and the estimated number of dwellings in the implementation 436.4. Average dwelling area of the test area classified by building’s year of

construction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 466.5. Comparison of the number of dwellings per building between the ob-

served values and the two estimations done in the implementation . 476.6. Number of inhabitants per building . . . . . . . . . . . . . . . . . . . 516.7. Households in the City of Berlin in 2011 by borough and household size 516.8. Comparison of the two methods for the estimation of the number of

inhabitants against the average number of resident per dwelling . . . 526.9. Building’s energy consumption results (kW/H Year) for different life

styles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 546.10. Average dwelling’s energy consumption results (kW/H Year) for dif-

ferent life styles at the surveyed buildings . . . . . . . . . . . . . . . 556.11. Average energy consumption per person results (kW/H Year) for dif-

ferent life styles at the surveyed buildings . . . . . . . . . . . . . . . 566.12. Case of study. Building parameters results . . . . . . . . . . . . . . . 606.13. Case of study. Food preparation energy appliance results . . . . . . . 616.14. Case of study. Laundry energy appliance results . . . . . . . . . . . . 616.15. Case of study. Electrical lighting energy appliance results. Business day 61


Beneficiario de Colfuturo 2010 vi



List of Tables

6.16. Case of study. Electrical lighting energy appliance results. Weekendday . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

6.17. Case of study. Entertainment and Technology energy appliance re-sults. Business day . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

6.18. Case of study. Entertainment and Technology energy appliance re-sults. Weekend day . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

6.19. Case of study. Personal Computer and Home Office energy applianceresults. Business day . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

6.20. Case of study. Personal Computer and Home Office energy applianceresults. Weekend day . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

6.21. Case of study. Miscellaneous energy appliance results. Business day . 636.22. Case of study. Miscellaneous energy appliance results. Weekend day 636.23. Case of study. Electrical energy consumption of a dwelling specified

by day, week and month . . . . . . . . . . . . . . . . . . . . . . . . . 646.24. Case of study. Annual values of electrical energy consumption per

dwelling, building and person . . . . . . . . . . . . . . . . . . . . . . 64

B.1. Example of the survey made to the residents . . . . . . . . . . . . . iii


Beneficiario de Colfuturo 2010 vii



Abstract

Abstract

In this master thesis, a methodology for the estimation of electrical energy demandof buildings using a 3D city model is presented. The idea of include that kind ofdata lies in the fact that nowadays those models are rich of geometric and semanti-cal data, which could be useful in energy demand forecasting.

The first part of the document introduces the basic concepts of the GeographicInformation Systems, the CityGML standard and the Web Map Services. As well asthe introduction of basic concepts of energy and the presentation of several methodsfor the estimation of electrical energy demand. After the presentation of the differentmethodologies, the selection of the End-Use method is done, which is the one thatbetter suits the scope of this master thesis. After the selection of the method to use,a better presentation of its characteristics is done.

Later the comprehensive tools for processing CityGML files are presented, givinga description of their characteristics and why are they considered. After this chapterthe problem statement of the master thesis is done.

In the following chapter the methodology for the estimation of electrical energydemand is done, first from a theoretical perspective being detailed in every singlestep of the process at the end the chapter is concluded by the presentation of thework-flow diagram that better states the ideas presented. After this chapter the re-sults of an area of study used in the implementation part of the master thesis areshown, in this section the methodology is evaluated and some adjustments to theinitial idea are done.

The final chapter of this document is dedicated to the conclusions and the pre-sentation of further researches that could take place after.


Beneficiario de Colfuturo 2010 viii



Zusammenfassung

Zusammenfassung

In dieser Masterarbeit wird eine Methodik fur die Prognose vom Strombedarf derGebaude mit der Benutzung eines 3D-Stadtmodell vorgestellt. Die Idee liegt umdiese Art von Daten zu nutzen, dass heute diese Modelle reichen von geometrischenund semantische Daten, die nutzlich sein in Energie Bedarfsprognose konnten.

Der erste Teil des Dokuments wird die grundlegenden Konzepte der Geographic In-formation Systems, der CityGML-Standard und den Web Map Services vorgestellt.Neben der Einfuhrung der grundlegenden Konzepte von Energie und die Auffuhrungvon verschiedene Methoden fur die Prognose des Strombedarfs. Nach der Prasentationder verschiedenen Methoden, die Auswahl der End-Use-Methode durchgefuhrt wird,das ist der eine, die passt besser zu das Ziel dieser Masterarbeit. Nach der Auswahlder Methode zu verwenden, wird eine bessere Prasentation seiner Eigenschaftengetan.

Im folgenden Kapitel wird die Methodik fur die Prognose des Strombedarfs erfolgt,zunachst ist aus theoretischer Perspektive jeder einzelne Schritt des Prozesses detail-lierte , und am Ender des Kapitels stellt die Prasentation der Work-Flow Diagramms,die besser den Prozess fest erklaren.

Nach diesem Kapitel werden die Ergebnisse eines Test Region die in der Umset-zung Teil der Masterarbeit verwendet gezeigt, in diesem Abschnitt ist die Methodikausgewertet und einige Anpassungen bei der ersten Idee sind fertig.

Das letzte Kapitel dieses Dokuments wird auf die Schlussfolgerungen und die Prasentationvon weiteren Forschungsthemen fokussiert, die in kommenden Projekten konntenausgebildet.


Beneficiario de Colfuturo 2010 ix



Acknowledgements

Acknowledgements

“ I can do everything by the power of Christ.He gives me strength”

Philippians 4:13

First of all I want to thank God for giving me the chance of making my dream cometrue, I could live and study in Berlin. I would like to thank Prof. Dr.-Ing. FrankNeitzel and Dipl.-Ing. Thomas Becker for being my tutors during my master thesisbut most, for their cooperation during the last months. I would like to thank as wellto the staff members of the Institute for Geodesy and Geoinformation Science whoalways are there ready to help us. Finally, to Prof. Dr. rer. nat. Thomas H. Kolbefor giving me the idea of working in this topic.

I would like to thank COLFUTURO, without their financial support none of thiswould ever happen, thanks for thinking that education is the way Colombia can havea better future.

To my family, my parents Pedro and Gladys for their love, prays and support,my sister Laura and my brother Diego and his family, who always were there tryingto cheer me up. To my little nieces Marıa Jose and Salome with their skype calls,Whatsapp messages, photos, letters and all those things they did to show me theirlove.

To Felipe, Ivan and Laura, my friends and my family all this years here in Berlin,for their help during my studies and their support when it was needed, you mademy time here happier. To Camilo “Tocayo”, for showing me the real meaning of theword friendship. And to Luz Adriana, for her unstoppable support.

Esto es por ti abuelito Efraın, siempre estaras en mi corazon.


Beneficiario de Colfuturo 2010 x



1. Introduction

1. Introduction

1.1. Motivation

At the present time, new technologies are developed every single day in such a va-riety of topics that could go from the wide popular internet and smart phones upto electric cars. Almost all of them share a factor in common, they need electricalenergy. This incredible request for electrical energy and the increase in the concernfor the environment lead the decision makers to demand a good knowledge of theelectrical energy consumption of a place (city, region, country). For that reason,Several approaches have been developed for the estimation of energy consumptionwith different scopes such as energy forecasting or the influence of human behaviourto mention some.

It is likely that most of those studies took place on countries located on high lat-itudes where seasons entail a dramatic change of temperatures. These researcheshave considered different methodologies to obtain their results, such as econometric,trends, inventories, end-user models and so on, but at this moment in time, noneof those that have been found and used as references for the elaboration of thisdocument do include 3D city models as an input parameter for the estimation ofelectrical energy consumption.

1.2. Problem Statement

Chapters 2 and 3 of this Master Thesis give us the theoretical background fromboth the CityGML standard and the forecasting of Electrical Energy Demand. Nomatter what estimation method is used, this is a meticulous and rigorous processthat requires detailed data, which must be included into the building simulationresulting in a more realistic assessment of energy demands (Yamaguchi, Fujimoto &Shimoda 2011, Lee, Yi & Malkawi 2011).

For this master thesis, the End-Use method is considered (section 3.1.5) becauseit focuses only on the impact of energy usage patterns of various devices and sys-tems. This is called Energy Appliances (Mehra & Bharadwaj 2000) and involvingfactors such as number of devices, number of users, time of use and its energyconsumption. None of those variables are external such as electrical energy prices,population income, social status, etc. Instead of that they are focused on facts thatcan be influenced by the building itself, for example the number of users of a mi-crowave oven for a 1 person flat will differ from a 4 people house.


Beneficiario de Colfuturo 2010 1



1. Introduction

It is expected that a well defined and detailed, from both geometric and seman-tic perspectives, 3D building model will lead to a better analysis of the ElectricalEnergy Appliances of a residential unit. Knowing the spaces of a building or anapartment influences directly on equipment requirements leading to a better defini-tion of the energy appliances database, i.e. a kitchen requires stoves, ovens, etc. Forthat reason this master thesis is developed, with the hope that the estimation theelectrical energy consumption of a residential building can be done based on its 3DCityGML model.





2. Geographic Information Systems


The first Geographic Information System GIS appeared at late 1960’s as an ini-tiative of the Canadian Government to produce Land capability maps. Those mapswere later analysed in order to obtain information of agricultural rehabilitation ofmarginal farms(Coppock & Rhind 1991), this system is known as the Canada Ge-ographic Information System (CGIS). Since their origin, GISs have been used ascomputer based systems for geographic information that allows the user to processlarge amounts of data, which is spatially located in an specific place.

Evolution of technology have entailed a benefit for the GIS sector. Now it is possi-ble to store massive amounts of data in a single computer. Data can be processedlocally as well in a remote way over a local network or as nowadays over the cloudeven with the use of portable devices such as mobile phones. Former time data wasmostly available in 2D, the initial results of a GIS were printed maps or tables thatwere used by the decision makers to support their statements. This initial concepthave changed by the time, the digital era offers multiple possibilities for the users tovisualise and present their results, an additional alternative are Web services, suchas Web Map Service (WMS). Based on dynamic user requests, the service returns araster version of a map i.e. a png file (de la Beaujardiere 2006). A second possibilityis a Web Feature Service (WFS), a type of service that from a user request returnsthe features or data itself that could even be modified locally by the user and afterstored again of the server (Vretanos Panagiotis 2005).

Nowadays most of the data and information available is on 3D, with new paradigmsof data acquisition and processing i.e. LIDAR, Digital Photogrammetry, Radar Satel-lite sensors. This kind of data contains not only x and y coordinates but also heightvalues (z coordinate) so new possibilities have emerged so users not only repre-sent the landscape with Digital Terrain Models (DTM) or Digital Elevation Models(DEM) but also to model human built structures like buildings, bridges, tunnels,etc., this kind of data has been modelled for many years using CAD software such asAutocad, Microstation or ArchiCAD. However new standards are requested so thedata can be read, shared and used between multiple users. One of them is CityGML(section 2.1) its functionality is much wider than just semantical and geometrical 3Dobjects representation. The CityGML can also be used in projects like (Lee 2004), tomodel Human Activity by using 3D GIS or as as data source for indoor navigationpurposes like in (Nagel, Becker, Kaden, Li, Lee & Kolbe 2010).

2.1. City Geography Markup Language

The City Geography Markup Language has been developed since 2002 as a part ofthe initiative Spatial Data Infrastructure Germany (GDI-DE), in 2008 version 1.0.0






was adopted as an Open Geospatial Consirtium (OGC) standard, being updated in2012 by CityGML 2.0 a major revision of the model, which introduces a “substan-tial addition and new features to the thematic model of CityGML” (Kolbe, Groger,Nagel & Hafele 2012). The standard was implemented as an application schema ofthe Geography Mark-up Language version 3.1.1 (GML3) based on the ISO 19107model.

This standard models 3D vector data with their associated semantic information,(Kolbe et al. 2012) it also provides an extension mechanism to enrich the data withidentifiable features under the preservation of semantic interoperability. CityGMLpresents a multi-scale model with 5 well-defined consecutive Levels of Detail (LOD),where objects are more detailed with increasing the LOD regarding both their ge-ometry and thematic differentiation. A CityGML file can contain multiple represen-tations and geometries for each object in different LOD simultaneously. Table 2.1gives description of the different Levels of Detail.

LOD0 LOD1 LOD2 LOD3 LOD4

Model scaledescription

regional,landscape

city, regioncity,city districts,projects

city districts,architecturalmodels (exterior),landmark

architecturalmodels(interior),landmark

Class of accuracy lowest low middle high very high

Absolute 3Dpoint accuracy(position/height)

lower thanLOD1

5/5m 2/2m 0.5/0.5m 0.2/0.2m

Generalisationmaximalgeneralisation

object blocksas generalisedfeatures;> 6 ∗ 6m/3m

objects asgeneralisedfeatures;> 4 ∗ 4m/2m

object as realfeatures;> 2 ∗ 2m/1m

constructiveelements andopenings arerepresented

Buildinginstallations

no no yesrepresentativeexterior features

real objectform

Roof structure/representation

yes flatdifferentiatedroofstructures

real objectform

real objectform

Roof overhangingparts

yes no yes, if known yes yes

CityFurniture noimportantobjects

prototypes,generalisedobjects

real objectform

real objectform

Solitary VegetationObject

noimportantobjects

prototypes,higher 6m

prototypes,higher 2m

prototypes,real objectform

Plant Cover no > 50 ∗ 50m > 5 ∗ 5m < LOD2 < LOD2

Table 2.1.: LOD 0-4 of CityGML with their proposed accuracy requirementssources: (Kolbe et al. 2012)

A visual example of the Levels of Detail can be seen in figure 2.1.






Figure 2.1.: The five levels of detail (LOD) defined by CityGMLsource: OGC City Geography Markup Language (CityGML) Encoding Standard (Kolbe et al. 2012)

As mentioned at (Kolbe et al. 2012), CityGML includes both spatial and thematicmodels (Section 2.1.1). The spatial model allows the consistent and homogeneousdefinition of geometrical and topological properties of CityGML features, represent-ing them as GML3’s geometry model objects, involving the use of their geometricprimitives. This means that the model is based on the standard ISO 19107 ”Spa-tial Schema” representing 3D geometry according to the Boundary Representationmodel.

A further definition of the spatial model can be found in the CityGML specificationdocument (Kolbe et al. 2012) section 8. The thematic model employs the geometrymodel for different thematic fields. Within this master thesis the Building Model, isof a special interest and it is explained at section 2.1.1. The standard offers also apossibility to model objects that are not explicitly modelled yet by using the conceptof generic objects and attributes (section 2.1.2).

2.1.1. Thematic Model

The thematic model of CityGML allows an explicit modelling of certain type ofobjects in order to reach a high degree of semantic definition and interoperabilitybetween applications. Most of the classes are derived from the basic geometric classesFeature and FeatureCollection for the representation of the spatial objects. How-

ever features can also contain non-spatial attributes which are mapped to GML3feature properties with their corresponding data types. This guarantees that at-tributes and data types will have a standardised interpretation.

The CityGML standard includes the following thematic extension modules: Appear-ance, Building, CityFurniture, CityObjectGroup, Generics, LandUse, Relief, Trans-






portation, Vegetation, WaterBody, and TexturedSurface. Thematic extension mod-ules Bridge, Tunnel are introduced in version 2.0 (Kolbe et al. 2012)

Building Model

According to (IGGS 2009a, Kolbe et al. 2012), this thematic model allows Buildingsto be represented in all levels of detail (LoD0 to LoD4). It enables the representationof simple buildings that consist of only one component as well as the representationof complex relations between parts of a building, e.g. a building consisting of threeparts a main house, a garage and an extension, those parts can again consist of parts.

An example of a simple and complex building can be seen in figure 2.2.

Figure 2.2.: Example of buildings consisting of building partssource: 3D Geo Database for CityGML (IGGS 2009a)

The subclasses ≪Building≫ and ≪BuildingPart≫ of ≪AbstractBuilding≫ enable thesemodelling options. The first two subclasses inherit all properties from the latter onelike its function, usage, year of construction, etc. (figure 2.3).






Figure 2.3.: Excerpt from the UML Diagram of the Thematic Building Modelsource: 3D Geo Database for CityGML (IGGS 2009a)

Attribute values are generally filled in the lower hierarchy level, because basicallyevery part can have its own construction year and function. However, the functioncan also be defined in the root of the hierarchy and therefore span over the wholebuilding. It is important to mention that the individual Building Parts within aBuilding must not penetrate each other and must form a coherent object.

Figure 2.4 presents the different levels of details LoD of CityGML. In LoD0 thebuilding is represented by horizontal surfaces describing the footprint and the roofedge. In LoD1, a building model consists of a geometric representation of the buildingvolume. This geometric representation is refined in LoD2 by additional MultiSurfaceand MultiCurve geometries, used for modelling architectural details like a roof over-hang, columns, or antennas. In LoD2 and higher LoDs the outer facade of a buildingcan also be differentiated semantically. Closure surfaces can be used to virtually sealopen buildings as for example hangars, allowing e.g. volume calculation.In LoD3,the openings in BoundarySurface objects (doors and windows) can be representedas thematic objects. In LoD4, the highest level of resolution, also the interior of abuilding, composed of several rooms, is represented in the building model by theclass Room.






Figure 2.4.: Levels of Detail of the Building Model of CityGMLsource: (Eicker, Nouvel, Schulte, Schumacher & Coors 2012, Kolbe et al. 2012)

2.1.2. Extending CityGML

The concept of generic objects and attributes allows for the extension of CityGMLapplications during runtime, i.e. any CityObject may be augmented by additionalattributes, whose names, data types, and values can be provided by a running ap-plication without any change of the CityGML XML schema. Similarly, features notrepresented by the predefined thematic classes of the CityGML data model may bemodelled and exchanged using generic objects.

In addition, extensions to the CityGML data model applying to specific applica-tion fields can be realised using the Application Domain Extensions (ADE) (Kolbeet al. 2012). Such additions comprise the introduction of new properties to existingCityGML classes like e.g. the number of habitants of a building or the definition ofnew object types. The difference between ADEs and generic objects and attributesis, that an ADE has to be defined in an extra XML schema definition file with itsown name space. This file has to explicitly import the XML Schema definition ofthe extended CityGML modules.






2.2. 3D-Geo-Database for CityGML

The Berlin 3D City Geodatabase was created on behalf of the Berlin Senate andBerlin Partner GmbH and funded by the European Union in the Period from Novem-ber 2003 and December 2005. It was developed at the Institute of Geodesy andGeoinformation Science at the TU Berlin extending the work done at the Institutefor Cartography and Geoinformation (IKG) of the University of Bonn and was imple-mented and evaluated in cooperation of Autodesk GmbH in Potsdam (IGGS 2009a).The database fulfils the CityGML standard model and was implemented in theDBMS Oracle 10G R2. In March 2012 a new version that fully supports the DBMSOracle 11G was released (IGGS 2012). This database fulfils the CityGML standardfor read and write files (versions 1.0.0 and 0.4.0.) and read (versions 0.3.0 and 0.3.1).

Being a compliant CityGML database means that it defines its classes and rela-tions for topographic objects based on important properties like geometry, topology,semantic and appearance (IGGS 2011). This data model supports up to 5 Levels ofDetail (LoD) for any geographical object (geo-object) meaning that when the LoDis increased, object obtain a more precise and finer geometry as well as its thematicrefinement. In the case of DTMs, it supports several types of representation andcould combine them when it is required. This also means spatial data types that arenot explicitly modelled in CityGML (raster files), for that purpose and in order tomanage orthophotos or aerial photographs, the Oracle 10G R2 GeoRaster function-ality is used.

The database model supports the storage and management of any geo-object thatis not already specified in CityGML. Nevertheless, it requires that those applicationsystems that would use these data must be able to interpret those file formats andafterwards retrieve them back to the database. Those objects could have externalreferences, which may help them to have a better definition or to extend the infor-mation about it e.g. a building with an id of a cadastre information system. Thereis also no restriction on the geometry of the 3D objects, they can be represented asthe combination of solids and surfaces and an aggregation of these elements.

This is an Open Source project and the entire software development is freely ac-cessible to the interesting public. (IGGS 2011, IGGS 2009a)





3. Energy

3. Energy

Generation and utilisation of energy involves a conversion from one energy forminto another, which in several situations requires intermediate steps. Energy couldbe classified (FAO. 1991) in primary, secondary, final and useful. Table 3.1 presentsthis classification, as well as some technologies and examples.

Energy technology examples

Primary coal, wood, hydro, dung, oil, etc.

conversion power plant, kiln, refinery, digester

Secondary refined oil, electricity, biogas

transport / transmission trucks, pipes, wires

Final diesel oil, charcoal, electricity, biogas

conversion motors, heaters, stoves

Useful shaft power, heat

Table 3.1.: Energy flowsource: Energy for sustainable rural development projects - Vol.1: A reader(FAO. 1991)

Where:

• Primary Energy: is the energy as it is available in the natural environment,i.e. the primary source of energy.

• Secondary Energy : is the energy ready for transport or transmission.

• Final Energy : is the energy which the consumer buys or receives.

• Useful Energy : is the energy which is an input into an end-use application.

The electrical energy, which is of relevance for the scope of this master thesis, ac-cording to (Princenton University 2012) is defined as ”energy made available by theflow of electric charge through a conductor” and as such the higher the voltage themore electrical energy is made available. It has multiple users and purposes in thecase of Domestic usage. It is employed for cooking, lighting, heating, ventilation andcooling (HAVC), communications, etc.

According to the SI (Organisation Intergouvernementale de la Convention du Metre2006), the measure unit for energy is the joule J, which can be expressed as the forceof one Newton applied on a distance of one meter, as can be seen in equation 3.1:

J =kg ∗ m2

s2= N ∗ m (3.1)

Where:





3. Energy

• J = Joule

• kg = kilogram

• m = meter

• s = second

• N = Newton

Power is the energy produced per unit of time (FAO. 1991), and electric power isthe rate at which electric energy is transferred by an electric circuit. The SI unit ofpower is watt, which is a joule per second and is expressed by equation 3.2:

W =m2 ∗ kg

s3=

N ∗ m

s=

J

s(3.2)

Energy consumption is measured by kilowatts per hour kWh which is the energyof one kilowatt power flowing for one hour(Silverman 2007), it can be expressed interms of Joules as:

1(kWh) = 3.6 ∗ 106J = 3.6 million Joules (3.3)

3.1. Methodologies used for Estimation of Electrical Energy

Demand

There are several public and private establishments in the energy sector that doelectricity forecasting as a fundamental part of their institutional mission. Estima-tions are done based on the data availability, scope or level of detail. In this section,some of the methods that have been used for forecasting are presented, followed bya evaluation of their pros and contras keeping in mind that the scope of this masterthesis is to use a city 3D model as a main source of data. At the end of this section,the selected method is presented.

3.1.1. Trend Method

In this method, the variable to estimate is considered as a function of time (Cullen1999, Mehra & Bharadwaj 2000), meaning that it determines the electricity demandas a trend of the historical sales of energy measured in kWh. Its basic approachconsiders the consumption of electricity for a sample year plus an additional amountfor each year after the base one. The calculation of electricity forecasting is doneusing the equation (3.4) from (Mitchell, Ross, Park & Corporation 1986):

(class kWh)year = a + b ∗ (year − base year) (3.4)

Where:

• a = estimate amount on a base year

• b = additional amount for each year after the base year

This was the leading method before 1970’s, however at the present time it is mainlyused to obtain a preliminary estimation of the forecasting or for short-term mod-elling. (Cullen 1999, Mitchell et al. 1986, Mehra & Bharadwaj 2000).





3. Energy

3.1.2. Time-Series Method

Other method that considers the variable as a function of time is the Time-SeriesMethod, based on (Cryer & Chan 2008) time series analysis purpose is both modelthe stochastic mechanism of an observed series and to forecast their future values.This method forecasts energy demand based on the patterns and trends found onthe data (Cullen 1999) which means that it requires electricity consumption valuesfor at least 30 time periods, i.e. months or years. This prerequisite might be a bigissue for regions with a lack of such a database as for example new developed areasin a city.

3.1.3. End-Use Method

A different practice can be done by the end use method, according to (Mitchellet al. 1986), this model is confined largely to forecasting residential loads, it surveysmajor electricity consuming residential equipment. Forecast is done by projectingquantity, energy efficiency and use of all electrical appliances used in a home. At theend, the final energy forecast results from the sum of all end-using activities (Mehra& Bharadwaj 2000). Equation 3.5 presents the estimation for electrical appliance.

ECA = S ∗ N ∗ P ∗ H (3.5)

Where:

• ECA = Energy consumption of an appliance in kWh

• S = Number of appliances per customer

• N = Number of customers

• P = Power required by the appliance in kWh

• H = Hours of appliance use

3.1.4. Econometric Method

”This approach combines economic theory with statistical methods to produce a sys-tem of equations for forecasting energy demand” (Mehra & Bharadwaj 2000). Foreach consumer class (residential, commercial, industrial, etc.) it estimates the en-ergy demand by the inclusion of the relationship between consumption with severalindependent factors such as economic, environmental, demographic, policy change,technological, etc. (Shuvra, Rahman, Ali & Khan 2011). It can project future val-ues for those factors and solve equations for future values of consumption, (Mitchellet al. 1986, Cullen 1999).

This is among the most complex forms of energy forecasting, and is used for allareas of service. In general, within this approach the estimation of energy demandcould be expressed by the equation 3.6 as:

E = f (Y, Pi, Pj , POP, T ) (3.6)

Where:





3. Energy

• E =Electricity demand

• Y = Income per capita

• Pi = Price of energy

• Pj = Price of related fuels (alternative fuel sources of energy if that’s the case)

• POP = Population

• T =Technology

Based on (Bhattacharyya & Timilsina 2009), “the following equations provide ex-amples of specifications used in simple econometric analyses” (table 3.2).

Where EMP is employment of labour, a, b, c, d, e, f, - are coefficients to be determined throughthe estimation process, t is time period t while t-1 represents the time period before t.(a) Linear relation between energy and income (GDP)Et = a + bYt

This implies an (income) elasticity that tends asymptotically to unity as income increases. Notethat b is not the elasticity in this specification, which has to be determined from the basic definitionof elasticity.(b) Log-linear specification of income and energyln Et = ln a + b ln Yt

Here b represents the elasticity of demand, which is a constant by specification.(c) Linear relation between energy and price and income variablesEt = a + bYt + cPt

This is not a popular specification however.(d) Log-linear specification of income, price and energyln Et = ln a + b ln Yt + c ln Pt

As with model (b), the short-run price and income elasticities are directly obtained here.(e) Dynamic version of log-linear specification of energy with price and income variablesln Et = ln a + b ln Yt + c ln Pt + d ln Et−1

The short run and long-run price and income elasticities are obtained here.(f) log-linear model of price and other demographic variablesln Et = ln a + b ln Pt + c ln EMPt + d ln P OPt

(g) log-linear model of energy, price, income, fuel share and economic structure variablesln Et = ln a + b ln Pt + c ln Yt + d ln Ft + e ln St

(h) dynamic version of the above modelln Et = ln a + b ln Pt + c ln Yt + d ln Ft + e ln St + f ln(Et−1)(i) linear relation between per capita energy and incomeEt/P OPt = a + bYt/P OPt

(j) Log linear relation between per capita energy and incomeln(Et/P OPt) = ln a + b ln(Yt/P OPt)(k) log-linear relation between energy intensity and other variablesln(Et/Yt) = ln a + b ln Pt + c ln Ft + d ln St

(l) Dynamic version of log-linear energy intensity relationln(Et/Yt) = ln a + b ln Pt + c ln Ft + d ln St + e ln(Et−1/Yt−1)

Table 3.2.: Typical examples of single equation econometric modelssource: (Bhattacharyya & Timilsina 2009)





3. Energy

3.1.5. Advantages and Disadvantages of the Mentioned Methods

Once the different forecasting methods are presented, it is necessary to discuss abouttheir benefits and finally to take a decision of which could be used on the estimationsthat are going to take place during this master thesis. For that reason Table 3.3 ispresented based on the consideration that a tabular presentation of their advantagesand disadvantages helps on the decision making.

Method Advantages Disadvantages

Trendmethod

-Little skill required-Inexpensive and quick-Can be upgraded by adjusting data-Minimal data requirements

-Vulnerable to changes-No explicit audit for errors

Time-Seriesmethod

-Accurate for short term-Low cost-Minimal data requirements-Statistical evaluation offorecast uncertainty

-Requires a big historical dataset-Does not treat factors explicitly-Tough interpretation of errors-Difficult to allow for conservationor change

End-Usemethod

-Can trace true location offorecasting error

-Intermediate technical andcomputer skills

-Easy to explain results

-Requires large amount ofdetailed data

-Data assembly costly and difficult-Technology must be explicitlyspecified

-Requires knowledge of end-usetechnologies and practices

Econometricmethod

-Explicitly measures effect ofunderlying causes of trendsand patterns

-Provides statistical evaluation offorecast uncertainty

-Combines economic anddemographic informationon service territory

-Can incorporate other methods-Models can be readily re-estimated

-Requires skill and experience ineconometrics and computerprograms

-Extensive data required for detaileddisaggregated model

-Costs can be relatively high

Table 3.3.: Advantages and disadvantages of the energy demand forecasting methodssources: (Mehra & Bharadwaj 2000, Mitchell et al. 1986, Cullen 1999)

Based on the motivation of this master thesis (section 1.1), and the informationpresented in this section, the End-Use method is chose as the Energy forecastingmethod that will be use in this master thesis. This method neither requires informa-tion as income per capita, energy price like the econometric method (section 3.1.4,or does not require an historical data base of the consumption values like Time seriesmethod requires (section 3.1.2).

Based on the fact that End-Use method focuses on the electrical energy appliancesfor its estimation, basic data such as the total number of dwellings and the numberof inhabitants per residential unit, must be available before the forecasting of theenergy demand of a residential building. Without those values the forecast can notbe done. Equation 3.5 presents that the number of users per appliance is one of itsvariables. Moreover the final value of the forecast of a dwelling is the sum of allelectrical appliances that have been modelled and the final value of the building’s





3. Energy

forecast is again the sum of all dwellings values plus those appliances of the buildingitself like for example electric light for common areas such as corridors.

Notwithstanding, it is important to emphasise that this method requires a detaileddatabase that includes as much appliances’ data as possible (i.e. Number of TV’s perhouse, energy requirements per TV, time that each TV is on per inhabitant, num-ber of inhabitants of that dwelling place, and so on). This kind of data is stronglyrelated to what will be called in this master thesis Occupancy influence, or whatsome authors (Page, Robinson & Scartezzini 2007, Shuvra et al. 2011, Santin, Itard& Visscher 2009) called Human Behaviour. This topic will be discussed in detail insection 3.2.

3.2. Occupancy Influence

Human behaviour has a huge impact on energy uses, it is relevant to energy useas mechanical parameters such as equipment and appliances (Santin et al. 2009)causing dramatic variations on energy consumption in similar dwellings with theidentical characteristics (Branco, Lachal, Gallinelli & Weber 2004). For that reasonit should not be undertaken during forecasting studies. Normally, it is included inenergy demand models of buildings as occupancy patterns that represent an aver-age user(Yamaguchi et al. 2011) or by operation schedules (Page et al. 2007, Leeet al. 2011). In case of the latter one, it is possible to assume that a continuouspresence of people at that location (dwelling or office) increases the energy use incomparison to those scenarios where users either do not have a constant presence orhave large periods of absence(Santin et al. 2009).

The electricity demand of a building (residential or offices) varies regarding to thenumber of people that lives or work in that specific place. An increase of the amountof users entails an increase of the energy demand in that location. Nevertheless thisincrease is not linear to the number of occupants(Haldi & Robinson 2010), and thathappens because some electrical appliances are shared between users without involv-ing a change on the energy consumption like the case of electrical lighting for a livingroom or cellular office. Some other appliances, which their use is shared most of thetime, only increases their consumption on specific circumstances like a TV, one ofthe users wants to see an specific show that nobody else is interested in. However,this statement is only applicable when all the occupants use the same device at thesame time. An additional TV or a washing machine, just to mention some devices,implies a high increase factor to the energy consumption of that specific appliance.

Another characteristic that has been distinguish in several approaches is users be-haviour (Reinhart 2004, Wout, Dirk & Hugo 2010, Astrid, Udo, Aris & Sabine 2010),those authors consider that users could be classified in active or passive. This is donein order to model dynamic switching of appliances based on external influences likefor example daylight can do to electrical lighting inside a building. Whereas a pas-sive user will turn the lights on at arrival and turn them off at departure, the activeone will be turning lights on and off during his stay according to daylight illuminance.





3. Energy

Due to the difficulty of modelling very dynamic and stochastic behaviours, someauthors like(Clevenger & Haymaker 2006, Haldi & Robinson 2010) have includedin their studies the human behaviour in extreme conditions, calling them as bestand worse cases. The best case is related to a normal employment of the electricalappliances by the user. The latter one means that all users are using all appliancesthe at the same time and all the time. Based on the paradigm presented within thisparagraph, several user profiles can be defined, which will be used as forecastingparameters in order to optimise results.

Figure 3.1 shows variability of normalised annual energy use for cold climates ob-tained by (Clevenger & Haymaker 2006), a study done to evaluate the impact ofbuilding occupant on energy modelling situations.

Figure 3.1.: Energy Results for Cold Climates. Example of Occupancy Influencesource: The Impact of the building occupant on energy modelling simulations (Clevenger & Haymaker 2006)

The blue line indicates the result obtained using the mean values of all parameters,variation of individual parameter values are presented in red (low use) or blue (highuse). The different behavioural scenarios tested in that research lead to variationsof more than 150% of the final energy use intensity.





3. Energy

3.3. Electrical Energy Appliances

After the presentation of several forecasting methods (see section 3.1), it is necessaryto specify the different energy appliances that will be considered during this masterthesis. However this method requires a big database in order to produce betterresults. At the end the electrical energy consumption of a dwelling can be expressedas the sum of its electrical appliances (equation 3.7).

DECi =n∑

j=1

ECAj + ǫj (3.7)

Where:

• DECi = Energy consumption for dwelling i as in equation 3.5

• ECAij = Energy consumption for end-use j, for dwelling i

• ǫj =Error term

There is an important consideration that must be include in equation 3.5. This isdue to the fact that a increase of the number of users do not mean a linear increaseof the energy consumption as it is expressed in (Haldi & Robinson 2010, Yamaguchiet al. 2011) and section 3.2. For that reason the number of users will be consideredon the mentioned equation as it is stated in (KEMA-XENERGY 2004, KEMA,Inc. 2010) so:

ECA = S ∗ U ∗ P ∗ H (3.8)

Where:

• S = Number of appliances per customer

• U = 1 + log N

• N = Number of users at that dwelling

• P = Power required by the appliance in kWh

• H = Hours of appliances use

By including the number of users just as its logarithm we are guaranteeing thatenergy consumption will increase for that energy appliance. It indicates that theincrement will be proportional to the number of users. The End-Use Energy GroupAppliances considered for this master thesis, which were defined based on (Meier,Rainer & Greenberg 1992, KEMA-XENERGY 2004, KEMA, Inc. 2010), are listedbelow.

• Food Preparation

• Laundry

• Electrical Lighting

• Entertainment and Technology

• Personal Computer and Home offices

• Miscellaneous





3. Energy

3.3.1. Food Preparation (ECFP)

This Appliance group include the following devices:

• Fridges and Freezers (ECFnF)A standard fridge/ freezer will be include per dwelling. This is one of thebasic electric devices that a dwelling has. Its energy consumption is obtainedfrom the specification document provided by the producer. This device willconsidered as on all the time. The energy consumption of these devices isobtained by the producer technical information sheets.

• Ranges, Ovens, Chimney hood, Microwave Ovens, Dishwasher (ECROCMD)Similarly, those appliances will be included per dwelling. Their energy con-sumptions are extracted from the specification document provided by the pro-ducer.

ECROCMD = U ∗ P ∗ H (3.9)

Equation 3.9 will be considered for each individual device. Finally, the electricalenergy consumption of the Food Preparation section can be specified as:

ECF P = ECFnF + ECR + ECO + ECCH + ECMO + ECD (3.10)

It is assumed that a dwelling has only one device per appliance.

3.3.2. Laundry (ECL)

This group appliance includes Washing machines and Tumble dryers, their energyconsumptions are obtained from the specification document provided by the pro-ducer. Due to their characteristics, they will be considered by their requirements ontime and energy consumption per load.

ECL = U ∗ P ∗ H (3.11)

Equation 3.11 will be considered for each individual device. The electrical energyconsumption of the Laundry appliance section can be specified as:

ECL = ECW M + ECT D (3.12)


3.3.3. Electrical Lighting (ECEL)

Considering the statement that electrical lighting requirements in a space type (i.e.Kitchen, bedroom, toilet, living room, etc.) are fixed (USA, Energy Dept. 2002),the estimation of electrical energy consumption will be done considering the listbelow:

• lamps per room LpR

• Rooms per dwelling RpD





3. Energy

• Operating hours per room HpR

• Wattage per Lamp WpL

The number of users per dwelling will be considered as it is stated in equation 3.8so the equation for the estimation of the Electrical Lighting per space type is:

ECEL = LpR ∗ RpD ∗ HpR ∗ WpL ∗ U (3.13)

3.3.4. Entertainment and Technology (ECEnT)

This group appliance include several devices such as Tv, video players (Blue Ray,DVD, VCR, etc.), Stereo, Home Theatre, Gaming Systems. Their energy consump-tions are obtained from the specification document provided by the producer.

ECEnT = U ∗ P ∗ H (3.14)

Equation 3.14 will be considered for each individual device. Based on this, the electri-cal energy consumption of the Entertainment and Technology sector will be specifiedas:

ECEnT = ECT V + ECV P + ECS + ECHT + ECGS (3.15)


3.3.5. Personal Computer and Home offices (ECPC)

This group appliance involves the number of personal computers as well as theirtime of use. There are specific devices that are considered being on all the timesuch as modems and routers, they do not require an specification of the number ofusers because that’s irrelevant for the equipment. A special case is printer devicesor scanners due to the fact they have such a dynamic schedule, a standard shortperiod of time will be considered. Their energy consumptions are obtained from thespecification document provided by the producer.

ECP C = U ∗ P ∗ H (3.16)

Equation 3.16 states the estimation of electrical energy consumption for computers,printers, scanners, etc. in case of modems and routers it is not necessary to includethe number of users for its estimation so as equation 3.17.

ECP C,MR = S ∗ P ∗ H (3.17)






3. Energy

3.3.6. Miscellaneous (ECM)

According to (Meier et al. 1992), Miscellaneous end use appliances are defined aselectricity not consumed by familiar end uses such as those mentioned above. Thisgroup has a high variability and it is highly influenced by the launch of new tech-nologies. Nowadays, mobile phones, tablets, docking stations, etc., can be includedinside this category. Their energy consumptions are obtained from the specificationdocument provided by the producer.

ECM = U ∗ P ∗ H (3.18)

Equation 3.18 will be considered for each individual device. It is assumed that adwelling has only one device per appliance.

3.3.7. Summary of Electric Model

In summary, the final electric model is expressed as the following equation:

DEC = ECF P + ECL + ECEL + ECEnT + ECP C + ECP C,MR + ECM (3.19)

In my opinion, all estimation must be adjusted to specific time frames such asdays, weeks or months, i.e. Laundry Appliances section 3.3.2 estimates its energyconsumption per load, it is necessary to specify the number of loads per week/monthso those values can be add to the estimation values obtained for other applianceswith other type time specifications such as fridges/freezers or wifi routers with theirdaily consumption (they are 24/7 on) or TV sets and their time of use per day. Itis important to state as well that the consumption of all energy appliances will beexpressed in kW/h.





4. Comprehensive tools for Processing CityGML Files

4. Comprehensive tools for Processing CityGML

Files

The conceptual models that have been presented in the previous chapters, will helpthe author with the development of this master thesis. Furthermore, they require aphysical implementation to take place and it will be done using a 3D City model,which is based on the CityGML standard. This require a complete implementationframework that involves the use of several technological tools. This will allow thedevelop and execute of the workflow of the master thesis and is described in chapter5.

4.1. citygml4j 2.0ea

It is a Java class library and API for facilitating work with the City GeographyMarkup Language (CityGML) developed at the Technical University of Berlin. Itallows to read, process, and write CityGML datasets, and to develop CityGML-aware software applications using Java (IGGS 2009b).

This master thesis employs this library for the estimation of energy consumptionof residential buildings. First of all, basic geometric and semantic information frombuildings is extracted including the estimation of the number of inhabitants. Afterthis process the electrical energy demand is computed. Final results are stored asgeneric attributes of buildings in a new CityGML File.

4.2. JTS Topology Suite 1.13

According to (Tsusiat Software 2011), “the JTS Topology Suite is an API of spatialpredicates and functions for processing geometry” using Java. It is a robust imple-mentation for processing linear geometry on 2D. It is of special relevance for thismaster thesis because all geometric enquires are solved using the library methodsand properties. GML geometries are converted into JTS geometries so propertiesor attributes of those geometries are easily to ask like for example the area of apolygon.

4.3. FME

The Feature Manipulation Engine developed by Safe Software is a collection of spa-tial ETL (Extract, transform and load) tools, which are useful in the process oftransforming and manipulating spatial data (Carrion 2010). Based on their own





4. Comprehensive tools for Processing CityGML Files

words (Safe Software Inc. 2012), it allows the transformation of 275 spatial and non-spatial formats, it is very useful to restructure, reformat and integrate spatial data.

This platform will be used to convert the CityGML file generated as a result ofthe processes done by the citygml4j library into other formats so those values areeasily readable and handle by other platforms. Format types like ESRI Geodatabaseor Excel table sheets are examples of data types that are supported by this plat-form.

4.4. ArcGIS

ArcGIS is a Geographic Information Software from ESRI. This platform is usedmainly for visualisation purposes in cases such as the presentation of the test areaor to present the resulting database using either ArcScene or ArcMap as well as forresult comparison purposes.





5. Estimation of Electrical Energy Demand using CityGML

5. Estimation of Electrical Energy Demand using

CityGML

This section presents the methodology done in this master thesis for the estimationof electrical energy demand of buildings using a 3D CityGML model. It presentsthe hypothesises and assumptions considered as well as and their theoretical justi-fication. At the end of each main part, it is resumed with a diagram that presentsthe workflow that is followed. The estimation is split into three main parts, first oneis the estimation of building parameters (including its number of inhabitants), thesecond part is energy appliances, it states which equipment are considered and thelast one is human behaviour, which presents the life style model used.

5.1. Building Data Requirements

A detailed 3D semantic model of a building is fundamental for the scope of thismaster thesis, it can provide relevant information such as number of storeys of abuilding, quantity of dwellings and their rooms. The CityGML standard providesthe right environment for those modelling purposes, table 5.1 shows the relation ofthe semantic themes that are available for the building model 2.1.1 and the level ofdetail representation.

Geometric / Semantic Theme LOD0 LOD1 LOD2 LOD3 LOD4

Building footprint and roof edge XVolume part of the Building shell X X X XSurface Part of the Building Shell X X X X

Terrain Intersection Curve X X X XCurve Part of the Building Shell X X X

Building Parts X X X XBoundary Surfaces X X X

Other Building Installations X X XOpenings X X

Rooms XInterior Building Installations X

Table 5.1.: Semantic themes of the class AbstractBuildingsource: OGC City Geography Markup Language (CityGML) Encoding Standard (Kolbe et al. 2012)

As the reader can see in table table 5.1, critical information such as volume, footprintor rooms are available when a building is modelled using this standard. Furthermore,table 5.2 presents the semantical attributes of a building that should be accessiblefrom the CityGML file i.e. number of storeys, storey height.






≪Feature ≫

AbstractBuilding+class : gml::CodeType [0..1]+function : gml::CodeType [0..*]+usage : gml::CodeType [0..*]+yearOfConstruction : xs::gYear [0..1]+yearOfDemolition : xs::gYear [0..1]+roofType : gml:CodeType [0..1]+measuredHeight : gml::LengthType [0..1]+storeysAboveGround : xs::nonNegativeInteger [0..1]+storeysBelowGround : xs::nonNegativeInteger [0..1]+storeyHeightsAboveGround : gml::MeasureOrNullListType [0..1]+storeyHeightsBelowGround : gml::MeasureOrNullListType [0..1]

Table 5.2.: AbstractBuilding Classsource: OGC City Geography Markup Language (CityGML) Encoding Standard (Kolbe et al. 2012)

However not all that information is accessible from the dataset available for thismaster thesis, this is shown in table 5.3, which presents the attributes of a building.

Name Type

EIG KL OV gen:intAttribute

EIG KL PV gen:intAttribute

ANZ LOC gen:intAttribute

EIG KL ST gen:intAttribute

FOLIE gen:stringAttribute

LFD gen:stringAttribute

HNR gen:stringAttribute

STR gen:stringAttribute

GMDE gen:stringAttribute

KREIS gen:stringAttribute

RBEZ gen:stringAttribute

LAND gen:stringAttribute

Kachel gen:stringAttribute

TexVersion gen:intAttribute

H Trauf Max gen:doubleAttribute

H Trauf Min gen:doubleAttribute

H First Max gen:doubleAttribute

H First Min gen:doubleAttribute

Table 5.3.: Attributes available at the CityGML dataset of Berlinsource: CityGML dataset file of the Test Area

The reader can see in table 5.3, that the dataset contains no storey informationas was stated by the AbstractBuilding class in table 5.2. Nevertheless the datathat is available is useful for the estimation of the missing information, which leadsto additional computations of the building features that are available when sucha detailed model is not present. This section presents the dataset of the area ofstudy and those concepts that are taken into consideration for the estimation of therequired data of buildings.

5.1.1. Test Area

A test area inside Berlin was defined for implementation purposes. The Import/Ex-port tool of the 3D CityDB (section 2.2) is used to extract the data of that area into






a single CityGML file, which contains a LOD2 geometric representation. This is thea fundamental task and so it can be seen what information is available from the 3DCity Model. Figure 5.1 shows the chosen test area that is located in the district ofCharlottenburg.

' OpenStreetMap (and) contributors, CC-BY-SA

Figure 5.1.: Test Areasource: Own graph using ArcGIS and OpenStreetMap as base Layer

The dataset includes 833 building type features.

5.1.2. Building Type

As it is mentioned in the problem statement, section 1.2, the scope of this masterthesis is estimate the electrical energy demand of residential buildings, which canbe classified by the CityGML function attribute, which contains the purpose of thefeature (IGGS 2009b). Due to the fact that we are focused on the Building ThematicModel, its function indicates whether it is either residential, public, industrial orcommercial building. Within the city model of Berlin, this attribute includes a codeaccording to the German cadastral information system (ALK) (Senatsverwaltungfur Stadtentwicklung 2005). Table 5.4 shows the function values that were identifiedas relevant, buildings which mainly or uniquely use is residential.






Function ALK Description

1210, 1211Wohnhaus in ReiheResidential House in row

1220, 1221Freistehender WohnblockFree standing block of flats

1230, 1231Wohnblock in geschlossener BauweiseBlock of flats in a closed construction

1300, 1301Gebaude und Freiflache - WohnenBuildings and open space - Living

1310, 1311 Einzelhaus Detached house1320, 1321 Doppelhaus Semi-detached house1330, 1331 Reihenhaus Serial House1340, 1341 Gruppenhaus Group home1360, 1361 Hochhaus high-rise

1381Behelfsmaßiges WohngebaudeProvisional residential building

1390Andere WohnanlageOther Condominium

1399WohngebaudeResidential building

2100Gebaude- und Freiflache - Mischnutzung mit WohnenBuildings and open space - mixed use with housing

2110Wohnen mit offentlichLiving with public

2120Wohnen mit Handel und DienstleistungenLiving with trade and services

2130Wohnen mit Gewerbe und IndustrieResidential, commercial and industrial

2140Offentlich mit WohnenPublic building with housing

2150Handel und Dienstleistungen mit WohnenTrade and services with living

2710 Wohnen Residential2711 Landwirtschaftliches Wohngebaude Farm house

Table 5.4.: Classification of residential buildings in the city model of Berlin accordingto the ALK

After an initial filter to the dataset, the result is that 553 of those features have afunction value that lies into the values presented on table 5.4.

5.1.3. Estimation of Number of Storeys

Table 5.3 indicates that the dataset do not have information regarding buildingstoreys, nevertheless this section presents an approach that uses the building bound-ary surfaces for that estimation. Based on the fact that the LOD2 representationmodels only the features that defines a building, as can be seen in figure 5.2(a), itis stated that a building is delimited at its nadir and zenith by the GroundSurfaceand RoofSurface features respectively, as well as the GroundSurface is its footprintor better said, its constructed area.

The statement made in the previous paragraph is to justify my decision of con-sidering only those parameters for the estimation of what I call the Living Spaceinside a Building, which is the space between the lower part of the Roof and thefootprint of the building as it is shown in figure 5.2(b).






RoofSurface

WallSurface

GroundSurface

(a)

Living Space

inside a Building

Attic

(b)

Figure 5.2.: LOD2 Building, all boundary surfaces Figure (a), delimitation of Build-ing’s Living Space, Figure (b).

source: Own graph using FZKViewer-2.3 (KIT 2012)

For that reason it is necessary to extract the limits of the building, which are theextreme coordinates of both features, in the case of the Ground Surface just onevalue for the Z coordinate is obtained, this is because the footprint of a building isa plane. For the Roof Surface, only the minimum Z coordinate value is used at thisstep so the height of the Living Space of a Building can be calculated by equation5.1.

BLSh = RSzmin − GSz (5.1)

The number of storeys of a Building can be estimated as the integer value of equation5.2

NoS = BLSh − SSh (5.2)

Where:

• NoS = Number of Storeys

• BLSh = Building’s Living Space height

• SSh = Standard Storey height

Based on (Neufert, Brockhaus, Kister, Lohmann & Merkel 2005), the StandardStorey height considered for this master thesis is 2,75m (figure 5.3). Nevertheless,it is important to state that the considerations and statements presented by theauthors of that book, are based according to the norms and standards valid at thatspecific period of time, which means that the Storey’s height can differ by the yearof construction erection.






Sto

rey

he

igh

t

0,00

+2,75

-25

-13

+2,69

+2,69

+2,75

in ground plans

Figure 5.3.: Height dimension in Sections and Elevationssource:Wohnungsbau-Normen: Normen, Verordnungen, Richtlinien. 25.... (Neufert et al. 2005)

For that reason a classification of buildings is done with the aim to obtain betterresults of the number of storeys. However this information is not available in thedataset of the area of study (Table 5.3). A possibility to solve this issue is usingone of the WMS that the Senate of Berlin has available on its FIS-Broker service((Senatsverwaltung fur Stadtentwicklung und Umwelt 2013)). That is the case ofthe Gebaude alte 1992-1993 service, which shows a scan image of a map of Berlinwith its buildings and their year of construction (figure 5.4).

(a) (b)

Figure 5.4.: WMS Berlin buildings’ age, area of study figure (a) legend of the WMSfigure (b)

source: Own graph using the FIS-Broker WMS (Senatsverwaltung fur Stadtentwicklung und Umwelt 2013)






Notwithstanding this WMS presents a classification of buildings (figure 5.4(b)),which means that buildings must be classified on those categories and the infor-mation that is available of the building’s storey height must be adjust to that classi-fication. Table 5.5 presents the year of construction and their corresponding storey’sheight values.

Year Height (m) Special Considerations

<1900 3,20

1900 - 1918 3,00

1919 - 19323,00 1st to 2nd Floors2,50

1933 - 19452,75 1st to 2nd Floors2,50

1946 - 19612,70 1st to 3rd Floors2,50

1962-1974 2,70

>1975 2,80

Table 5.5.: Height of Buildings based on the year of constructionOwn table based on the FIS-Broker WMS (Senatsverwaltung fur Stadtentwicklung und Umwelt 2013), (Architekten- und

Ingenieur-Verein zu Berlin 1974) and (IGGS 2013)

These values are used as an input in the implementation, for every building that donot have year of construction data uses the standard value mention in this section2,75m. It is important to mention that height values do not include the space betweenstoreys or the floor itself. This mean an additional 40 cm per storey.

Data Acquisition

The result from a WMS is a image, so addi-tional steps are required to extract this data. Although there are many possibilities to that automatic using platforms like FME, ERDAS Image, ArcGIS, etc., I decided to do it manually due to the image resolution of the WMS and overlapping mismatch (figure 5.5).

Figure 5.5.: Overlapping mismatch of WMS and buildings’ footprintsource: Own graph using ArcGIS



nana

Typewritten Text

nana

Typewritten Text

nana

Typewritten Text

nana

Sticky Note

Unmarked set by nana




The results of building classification of the area of study can be seen in table 5.6.

Year Number of buildings

<1900 239

1900 - 1918 82

1919 - 1932 0

1933 - 1945 1

1946 - 1961 127

1962 - 1974 71

1975 33

Table 5.6.: Classification of building age for the area of studysource: Own table

Roof Space

The last parameter that must be considered in this section is the roof space, there isno statement that says if the roof space is suitable for living or not. Notwithstandingat least in the case of Germany, several Federal Estates define on their legislationfor buildings that those areas can only be considered as suitable for residentialpurposes if they have a height value of at least 2, 30m as can be seen on (BayerischeStaatregierung 2007, Hamburg Senat 2005, Recht NRW 2000). For that reason Iconsider as an additional storey of the building if its roof has a height higher than2, 30m or better expressed.

Roofh = RSzmax − WSzmax (5.3)

Where:

• Roofh = Roof height

• RSzmax = max. Z coordinate of the Roof Surface Bounding Box

• WSzmax = max. Z coordinate of the Wall Surfaces

5.1.4. Public Areas Inside a Building

Once the number of storeys for each building is estimated, it is necessary to subtractfrom each storey the common or public areas, where its inhabitants can move insidethe building for either departure or arrival purposes. This consideration is donefor two reasons, the first one is that those areas require different Electrical EnergyAppliances and the second one is that after the removal of those areas, the remanningone can be assumed as private or area assigned to dwellings. This kind of assumptionswill not take place on the moment that a LOD4 of the building is available.






Figure 5.6.: Ground Surface feature of a buildingsource: Own graph using FZKViewer-2.3 (KIT 2012)

Figure 5.6 presents the GroundSurface of a building, as it was stated at section 5.1.3,this feature is considered as the footprint or constructed area of a building, examplesof area that can be removed from this feature is presented in figure 5.7 and its basedon (Neufert et al. 2005).

Figure 5.7.: Several designs of public stairs inside a buildingsource: Own graph based on (Neufert et al. 2005)

As it was stated before in this document, section 2.1.1 , a Building can be modelledas a complex relation of multiple Building Parts the Building’s footprint area will bethe sum of its GroundSurfaces. Based on (Loga, Diefenbach, Knissel & Born 2005),the 25% of the footprint of the building will be assigned as its common area.

5.1.5. Dwellings / Inhabitants

This is a critical part of the master thesis because the number of dwellings andinhabitants per storey are estimated. Several authors such as (Wohnungsanwalt 2009,Amt fur Statistik Berlin-Brandenburg 2011) and (DESTATIS 2009) indicate thatthe average Flat area in Berlin is 71m2, furthermore (Wohnungsanwalt 2009) statesthat rooms have an average size of 19, 57m2. An approach on the estimation ofthe number of inhabitants of a building could done based on (Senatsverwaltung furStadtentwicklung 2010) which states the following statistical values for Berlin.






• 70, 4m2 – Mean area per dwelling

• 38, 8m2 – Mean area per person in a inhabited dwelling

• 1, 82 – Mean number of people per dwelling

Another approach to the estimation of the number of dwellings in a building is theuse of several flat prototypes as they are presented in (Neufert et al. 2005). Someexamples are shown in both table 5.7 and figure 5.8.

Dwelling Area in m2

1-Person Flat 40,46

2-People Flat 56,47

3-People Flat 80,50

4-People Flat 103,23

Table 5.7.: Relation between the number of inhabitants per dwelling place and itscorresponding area in square meters

source: Bauentwurfslehre: Grundlagen, Normen, Vorschriften (Neufert et al. 2005)

The following examples shown a direct relation between the number of inhabitantsand the number of rooms in a dwelling. Those apartments present an increase oftheir size in m2 close to the value stated by (Wohnungsanwalt 2009) of 19, 57m2.

Bedroom

Entrance

Toilet

Living

room

(a)

Bedroom

Living

room

Kitchen

Toilet

(b)

Kitchen

Living

roomEating room

Bedroom

Bedroom

Toilet

Toilet

balcony

balcony

(c)

Figure 5.8.: Examples of dwellings, figure(a) one room flat (40m2), figure (b) tworooms flat (54m2), figure (c) three rooms flat (95m2)

source: Translate graphs based on (Neufert et al. 2005)

Those examples presented in figure 5.8 are only for visualisation purposes, no specialdwelling or building design is considered in this master thesis due to the complex-ity that it implies. The statistical values of(Senatsverwaltung fur Stadtentwicklung2010) will be the approach that is used for the estimation of the number of inhabi-tants of a building an consequently on the estimation of electrical energy demand.






5.1.6. Workflow of the Estimation of Building Parameters

Figure 5.9 presents the UML activity diagram1 for the estimation of building prop-erties in this master thesis.

Read

th

e C

ityG

ML

File

citig

ml4

j

<<

localP

reconditio

n>

>

CityG

ML F

ile

A

RoofS

urf

ace

Gro

undS

urf

ace

Read

Bo

un

dary

Su

rfaces

Geo

metr

y

Req

uest

Geo

metr

y

Req

uest

RoofS

urf

ace

JT

S T

opolo

gy

Gro

undS

urf

ace

JT

S T

opolo

gy

Bo

un

din

gB

ox

esti

mati

on

Bo

un

din

gB

ox

esti

mati

on

<<

localP

ostc

onditio

n>

>

<<

localP

ostc

onditio

n>

>

<<

localP

ostc

onditio

n>

>

<<

localP

reconditio

n>

>

esti

mati

on

of

Bu

ild

ing

Para

mete

rs

Bu

ild

ing

Year

of

Co

nstr

ucti

on

FIS

-Bro

ker

WM

S

Resid

ential B

uild

ing

YE

SN

O

NO

It is a

Build

ing

Figure 5.9.: Activity Diagram for the estimation of Building Parameterssource: Own diagram

1Unified Modelling Language. “shows the workflow from a start point to the finish point detailingthe many decision paths that exist in the progression of events contained in the activity” (SparxSystems Pty Ltd 2013)






5.2. Energy Appliances

As it is stated in section 3.1.5 the End-Use method is the selected one for the energyforecasting. Furthermore, section 3.3 presents the definition of the electrical energyappliances as well as indicates the devices taken into consideration. This sectionpresents the energy consumption values of that equipment.

Values present here we obtained from several web sources, the main one is a Website of the Ministerio de Energıa y Minas (Ministry of Energy and Mining) of Peruthat presents an explanation of the estimation of consumption of a residential client(a dwelling), including basic concepts and a list of house equipment with their con-sumption values (Min.EnergiaMinas n.d.). Those values are presented in table 5.8.

Appliance Consumption in kWhFridge 0.35 / per yearFreezer 0.25 / per year

Freezer / Fridge 0.408(1) / per yearHob (4 cooking places) 4.5

Ovens 1.56(1)

Chimney Hood 0.3Microwave Oven 1.2

Dishwasher 1.44(1)

Washing Machine 0.63(1) / per loadTumble Dryer 2.5 / per load

Electrical Lighting (Lamps)Lamp 100 0.1Lamp 40 0.04Lamp 32 0.032

Saverlamp 0.02

TV0.176 (6)

0.001 Standby

Home Theater 0.105 (5)

Gaming System0.002 Standby

0.102 onPersonal Computer 0.2

Printer/scanner 0.01Modem ADSL 0.03

Mobile Phone 0.005(4)

Tablet 0.012(4)

Docking station 0.007(3)

Vacuum 1.3Iron machine 1

Coffee machine 0.6Digital TV Adapter 0.01

Table 5.8.: Energy Appliances and their energy consumptionsources: (Min.EnergiaMinas n.d.), (1) (Carbon Footprint Ltd. n.d.), (2) (Striatum Energy 2010), (3) (Sony 2013), (4)

(Ho 2009), (5)(Philips 2013), (6)(Bluejay 2010) Value of a 42” LCD TV

This method requires the specification of the energy appliances that are consideredfor the forecasting as well as the number of units and users, this master thesis will






consider only one device per appliance. The number of users per dwelling will be theresult of the process done in section 5.1.

5.3. Human Behaviour

After the presentation of the energy appliances ( section 5.2). It is necessary to definea life style model or schedule of users as it is mention in (Yamaguchi et al. 2011). Forthat reason I divide the week on two type of days, weekday (monday to friday) andweekends (saturdays, sundays, public holiday). Days are split in three main sectors,sleeping time, at home and outside, and here is where the difference between daysis presented. This is possible to see in figure 5.10.

Figure 5.10.: Proposal of classification of Occupants behavioursource: Own graph

The purpose of splitting the day on three main activities is to propose a time frame-work that indicates when the residents are at home, during that time there is a cleardifference on energy consumption when people are either awake or sleeping.

After the definition of the general time framework, a time schedule is defined foreach of the appliances as is presented in (Clevenger & Haymaker 2006). This is donein order to analyse human occupancy influence and is classified in three Ranges low,

medium, high for each energy appliance (table 5.9).






Appliance High Medium Low

Food Preparation 2h 1h 0.5h

Fridge/Freezer on all time

Laundry 2 loads/week 1 load/week 0,5 load/week

Electrical Lithing 16h 8h 4h

Entertainment and Technology, Personal Computer Miscellaneous

Weekday 6h 3h 1, 5h

Weekend 8h 4h 2h

Microwave Oven 20min 10min 5min

Table 5.9.: Schedule ranges per energy appliancesource: Own table

The time of use of each energy appliance was defined only when people are awakewith the respective specific exception of fridges and modems/routers because theyare considered on all the time.

5.3.1. Workflow of the Estimation of Energy Consumption

Figure 5.11 presents the UML activity diagram for the estimation of energy consump-tion of a building. This figure is a continuation of figure 5.9 as building propertiesare required for this purpose.



Est

imat

ion

of

Elec

tr

icE

ner

gy

Dem

an

d

usin

g3D

Cit

yM

od

els

5.

Estim

atio

nof

Elec

trical

En

erg

yD

em

an

dusin

gC

ityG

ML

iterative

estimation Building

Parameters

Food Preparation

Laundry

Electrical Lighting

Entertainment and

Technology

Personal

Computer

Miscellaneous

Estimation of

Electrical Energy

Demand

<<localPrecondition>>

<<localPostcondition>>

Storage

of results

<<localPrecondition>>

File

Transformation

Analysis of

Results

A

Fig

ure

5.11.:A

ctivity

Dia

gramfor

the

estimation

ofE

nergy

Con

sum

ption

sou

rce:

Ow

ng

rap

h

Cam

iloA

lexan

der

Leon

San

chez

Ben

eficiario

de

Colfu

turo

201037



6. Results

6. Results

The work-flow presented in section 5 leads to the estimation of the energy demandof a building based on a LoD2 CityGML model which is the scope of this masterthesis (Section 1.2). Furthermore it leads as well to the calculation of some othervalues that can be obtained such as the number of storeys, dwellings and residents ofthose buildings. The following section presents those results and analysed them, firstthe building parameters values that are required for the estimation of the energydemand of a building (section 5.1) to conclude with the energy demand values.

6.1. Building Parameters

In order to check the results a small survey was done inside the area of study, theinformation of 144 buildings was collected. The survey involved to count the numberof storeys per building and the number of rings that were available at the entrancesof the buildings assuming that each one of them belong to a different flat. There wasno special criteria for the survey despite of just to walking through the borough.Figure 6.1 shows the residential buildings in the area of study, marked with bluecolour all those that were observed.

Figure 6.1.: Survey of the area of studysource: Own graph using ArcGIS





6. Results

Table 6.1 shows the classification of the surveyed buildings by their year of construc-tion as they are presented in table 5.5.

Year of Construction Number of buildings

<1900 63

1901 - 1918 24

1946 - 1961 37

1962 - 1974 12

>1975 8

Table 6.1.: Surveyed buildings classified by building’s year of constructionsource: Own graph

Figure 6.2 presents the frequency of the number of storeys per building groupingthem by their year of construction.

Figure 6.2.: Number of storeys per building classified by year of constructionsource: Own graph using Excel

6.1.1. Storeys per Building

Two methods were used for the estimation of building’s number of storeys, in thefirst one the standard Storey Height (SH) and the second one the storey heightclassified by year of construction (YoC) as it is explained in section 5.1.3. Table 6.2presents a small table of the results obtained.





6. Results

Building ID Storey SStd.H ǫ BldgAge SH,YoC ǫ

BLDG 0003000000067203 7 9 -2 1961 9 -2

BLDG 0003000000069010 6 9 -3 1899 7 -1

BLDG 000300000006f0e2 6 8 -2 1899 6 0

BLDG 000300000006e689 5 10 -5 1918 6 -1

BLDG 0003000f001d3542 6 7 -1 1961 7 -1

BLDG 0003000f001f67be 6 10 -4 1899 7 -1

BLDG 0003000e005b3d5a 7 7 0 1899 6 1

BLDG 0003000000071392 5 5 0 1975 5 0

BLDG 0003000f001d355a 5 6 -1 1961 6 -1

BLDG 0003000f001d35a9 6 7 -1 1961 7 -1

Table 6.2.: Comparison of number of storeys per building between the observed val-ues and the estimated number of storeys in the implementation

source: Own table

Table 6.2 shows that a discrimination of the year of construction and its respectivestandard storey height does have an massive impact on the results of the estimationof the number of storeys of a building. Errors between the observations and theestimated values are considerably reduced by including this variable in the estimationprocess. The standard deviation σ of both methods is presented below. SStd.H meansstandard storey height, SH,YoC for the storey height based on the building year ofconstruction.

• σSStd.H= 1.15265

• σSH,YoC=0.59115

The histograms of both estimation methods are shown in figure 6.3.

−5 −4 −3 −2 −1 0 10

10

20

30

40

50

60

Fre

qu

en

cy

Difference in number of storeys

(a)

−2 −1.5 −1 −0.5 0 0.5 10

10

20

30

40

50

60

70

80

90

100

Fre

qu

en

cy

Difference in number of storeys

(b)

Figure 6.3.: Histograms of the estimation of number of storeys per building, fig-ure(a) shows the histogram for the SStd.H method, figure (b) shows thehistogram for the SH,YoC

source: own graphs using Matlab





6. Results

Based on (Greenwood & Nikulin 1996), the statistic testx2 is done to evaluate thegoodness of fit of data with equation 6.1.

x2 =n∑

i=1

(Oi − Ei)2

Ei

(6.1)

Where:

• x2 = Cumulative statistical test, which asymptotically approaches a normaldistribution

• Oi = Observed value

• Ei = Expected value

• n = Number of observations

With a confidence level of 95% and a degree of freedom of 143, the critical valueis Tx2 = 171.907 so the HO is accepted if the statement shown in equation 6.2 isfulfilled.

qα,f < Tx2 (6.2)

The results obtained by both methods are presented below.

• SStd.H : 62.8764 < 171.907

• SH,YoC : 19.955 < 171.907

For both methods the null hypothesis is accepted meaning that the observationsfollow a normal distribution.

Despite of the fact that error between observations and measured values were min-imised, most of the buildings shown in table 6.2 have an error of one or two storeys.In my opinion, it is important to state that there is no physical parameter thatdefines if a Roof is used as an attic. Many buildings that physically could use thatspace as residential do not used it without a known reason. Examples of such a caseare presented in figure 6.4





6. Results

(a) (b)

(c)

Figure 6.4.: Example of buildings with and without without residential attics. Fig-ure(a) shows the location of the buildings, figure (b) shows a perspectiveimage of the buildings. Figure (c) shows a building with its attic underconstruction

source: own graphs using Apple Maps for iPad

Despite of the fact that the three buildings that are shown in figure 6.4(b), have thesame type of roof based on what is visible in the image, only the middle uses its roofarea for residential purposes (see the big windows and balconies in the roof) so it isan additional storey for that building. Figure 6.4(c) shows a building that was re-furbished and during this process its roof was renewed and turned into a residentialspace.

Another physical factor that could had influenced into a wrong estimation of thenumber of storeys is the additional constructions that are on the roofs, which couldbe seen on several buildings (figure 6.5). This ”additional” level or construction isdone so users or residents can go to the roof. Figure 6.5(a) shows an example of abuilding with a flat roof. Figure 6.5(b) shows an example of several building withdifferent kind of roofs and some additional constructions on them.





6. Results

(a) (b)

Figure 6.5.: Examples of additional roof constructions

source: own graphs using Apple Maps for iPad

6.1.2. Dwellings per Building

The statistical value stated in (Senatsverwaltung fur Stadtentwicklung 2010) (section5.1.5) about the average size of a flat in Berlin is used for the estimation of thenumber of dwellings per building, first the number per storey is calculated and thenmultiplies by the number of storeys to obtain the final value of the building. Table6.3 shows the comparison of results for the same buildings that were presented insection 6.1.1.

Building ID Dwellings DStd.H ǫStd.H BldgAge DH,YoC ǫH,YoC

BLDG 0003000000067203 36 63 -27 1961 63 -27

BLDG 0003000000069010 12 36 -24 1899 28 -16

BLDG 000300000006f0e2 12 48 -36 1899 36 -24

BLDG 000300000006e689 22 60 -38 1918 36 -14

BLDG 0003000f001d3542 18 21 -3 1961 21 -3

BLDG 0003000f001f67be 12 40 -28 1899 28 -16

BLDG 0003000e005b3d5a 13 21 -8 1899 18 -5

BLDG 0003000000071392 24 25 -1 1975 25 -1

BLDG 0003000f001d355a 10 12 -2 1961 12 -2

BLDG 0003000f001d35a9 18 21 -3 1961 21 -3

Table 6.3.: Comparison of number of dwellings per building between the observedvalues and the estimated number of dwellings in the implementation

source: Own table

Where:

• DStd.H : number of dwellings considering a standard storey height SH

• DH,YoC : number of dwellings using a different storey height based on thebuilding’s year of construction

The user can see that there are dramatic differences between observations and theresults obtained using the average size of a flat in Berlin, the first conclusion canbe the discrepancies on the values of the number of storeys per building. Anotherstatement that should be considered is that the value used might be the average size





6. Results

area of the hole Berlin but either it does not represent properly the area of studyor that value should be discriminated as well by the building year of construction.The standard deviation for both set of results are presented below as well of theirhistograms (figure 6.6).

−80 −70 −60 −50 −40 −30 −20 −10 0 100

5

10

15

20

25

30

35

40

45

Fre

qu

en

cy

Difference in number of dwellings

(a)

−50 −40 −30 −20 −10 0 10 200

5

10

15

20

25

30

35

40

Fre

qu

en

cy


(b)

Figure 6.6.: Histograms of the estimation of number of dwellings per building, fig-ure(a) shows the histogram for the SStd.H method, figure (b) shows thehistogram for the SH,YoC method

source: own graphs using Matlab

The statistical test Tx2 will be done, being H0 that the dataset follows a normaldistribution as it was done to the results in section 6.1.1. The standard deviation ofthe two methods are shown below.

• σDStd.H= 13.45284

• σDH,YoC= 9.25783

With a confidence level of 95% and a degree of freedom of 143, the critical valueis Tx2 = 171.907. H0 is accepted if equation 6.2 is fulfilled. The results of bothestimations are shown below.

• DStd.H : 1095.230 > 171.907

• DH,YoC : 2918.346 > 171.907

For both methods H0 is rejected, meaning that the datasets do not follow a normaldistribution.

Dwelling estimation by building’s year of construction

After the initial results, I decided to estimate the average dwelling size of the sur-veyed buildings for the year of construction ranges that were stated in table 5.5.The area per dwelling for the observed buildings was obtained by calculating thefull constructed area of the building (summing the area of all ground surfaces times





6. Results

its number of storeys) and then dividing that value by the number of dwellings ofthat building (equation 6.3).

Area =ConstructedArea · NoStoreys

NoDwellings(6.3)

Once that value was obtained, the average area per dwelling for the test areawas estimated as 82.34m2, which totally differs from Berlin’s average flat size of70.4m2(Senatsverwaltung fur Stadtentwicklung 2010). Figure 6.7 presents the scat-ter plot of dwellings’ areas for the surveyed buildings.

0 50 100 15020

40

60

80

100

120

140

160

180

200

Are

a p

er

Dw

ellin

g (

m2)

Building ID: BLDG_000300000006c2c4

Figure 6.7.: Scatter plot of dwelling’s areas at test area. The green line representsthe average size for the test area, the red line represents the average sizefor Berlin

source: Own graph

Due to the discrepancies of areas and the rejection of the null hypothesis, I decidedto estimate the average size per residential unit according to surveyed buildingsgrouping them by their year of construction with the expectation of improving theestimation results. To do so average dwelling areas per year range was estimatedusing the surveyed data. Figure 6.8 shows the scatter plots for the different groups.

0 10 20 30 40 50 60 7040

60

80

100

120

140

160

180

Are

a p

er

Dw

ell

ing

(m

2)

(a)

0 5 10 15 20 2520

40

60

80

100

120

140

160

Are

a p

er

Dw

ell

ing

(m

2)

(b)





6. Results

0 5 10 15 20 25 30 35 4040

50

60

70

80

90

100

110

120

Are

a p

er

Dw

ell

ing

(m

2)

(c)

0 2 4 6 8 10 1220

40

60

80

100

120

140

160

180

200

Are

a p

er

Dw

ell

ing

(m

2)

(d)

1 2 3 4 5 6 7 850

60

70

80

90

100

110

120

Are

a p

er

Dw

ell

ing

(m

2)

(e)

Figure 6.8.: Scatter plot of dwelling’s areas at test area according to its year ofConstruction (a) (< 1900), (b) (1900 − 1918), (c) (1946 − 1961), (d)(1962 − 1974), (e) (> 1975). The green line represents the average sizefor the test area in that year range and the red line represents theaverage size for Berlin

source: Own graph

Table 6.4 presents the average area value for each year of construction category.

Year of Construction Average Area

< 1900 91.11

1900 − 1918 89.32

(1946 − 1961) 62.31

(1962 − 1974) 85.68

(> 1975) 80.00

Table 6.4.: Average dwelling area of the test area classified by building’s year ofconstruction

source: Own table

A new estimation of the number of dwellings of a building was done based on thevalues presented in table 6.4. A small presentation of the results can be seen in table6.5.





6. Results

Building ID Dwellings DH,YoC ǫH,YoC BldgAge DHA,YoC ǫHA,YoC

BLDG 0003000000067203 36 63 -27 1961 54 -18

BLDG 0003000000069010 12 28 -16 1899 21 -9

BLDG 000300000006f0e2 12 36 -24 1899 24 -12

BLDG 000300000006e689 22 36 -14 1918 30 -8

BLDG 0003000f001d3542 18 21 -3 1961 21 -3

BLDG 0003000f001f67be 12 28 -16 1899 21 -9

BLDG 0003000e005b3d5a 13 18 -5 1899 12 1

BLDG 0003000000071392 24 25 -1 1975 20 4

BLDG 0003000f001d355a 10 12 -2 1961 12 -2

BLDG 0003000f001d35a9 18 21 -3 1961 21 -3

Table 6.5.: Comparison of the number of dwellings per building between the observedvalues and the two estimations done in the implementation

source: Own table

Where:

• DH,YoC : number of dwellings using a different storey height based on thebuilding’s year of construction

• DHA,YoC : number of dwellings using a different storey height and an averagedwelling size based on the building’s year of construction

Having different dwelling sizes based on the year of construction lead to a reductionon the discrepancies in the estimation, as can be seen on its histogram (figure 6.9).

−50 −40 −30 −20 −10 0 10 200

10

20

30

40

50

60

Fre

qu

en

cy


Figure 6.9.: Histogram of the number of dwellings considering an average size peryear of construction

source: Own graph

Finally the statistical test x2 is done to evaluate if data follow a normal distribution,when the H0 (equation 6.1) is fulfilled. The test is done with a confidence level of95% and a degree of freedom of 143, the critical value is Tx2 = 171.907. The resultsare shown below.

• σDHA,YoC= 7.367

• DHA,YoC : 356.6528 > 171.907





6. Results

The result indicates that the dataset does not follow a normal distribution, it couldbe that the dataset neither represent correctly the are of study of it is not significantand a new survey should be done.

In my opinion, there are three factors that could help to understand the results ob-tained. The first one is the methodology of the estimation of the number of dwellingsusing an average area size. That average number might be for the hole city of Berlinbut it clearly does not represent the borough where this master thesis was done. Forthat reason I decided to estimate an average dwelling size for the area and specifi-cally for the year of construction ranges.

The second factor is regarding the methodology used for the estimation. As it wasstated in section 5.6 the constructed area was estimated as the result of the sum allGroundSurface features, because a building can be modelled as a complex relationof multiple building parts. Even though the model does split buildings into multipleparts, it is not explicitly expressed as it is declared by the encoding standard (figure6.10).

Figure 6.10.: Extract of the CityGML encoding standard for the declaration of build-ing parts

source: Own graph

This means that it is not possible to identify the different parts of a building in theimplementation so the ground surface is always considered as a hole and buildingheight is not discriminated for each part. For understanding purposes figure 6.11shows some examples of buildings that were modelled with a complex relation ofbuilding parts but the CityGML file does not explicitly present them. Figure 6.11(a)shows the CityGML model of a building in the area of study. Figure 6.11(b) showsthe same building in 3D as well as a sketch of the building as it was read in theimplementation (read box).





6. Results

(a) (b)

Figure 6.11.: Example of buildings with multiple usessource: Own graph

The last factor that is the buildings that were included as residential during thismaster thesis. Some of those buildings have a function code that states multiplepurposes, i.e. 2100 “Building and open space - mixed use with housing” (table 5.4).Meaning that part of the building or complete storeys are not residential so for thatreason the total number of dwellings is lower than the author would expected insome of them, for instance a 5 storeys building, like the one shown in figure 6.12(a),have 8 residential units instead of 10. Figure 6.12 presents some photos that werecaptured during the survey.

(a) (b)

Figure 6.12.: Example of buildings with multiple usessource: Own graph





6. Results

6.1.3. Number of Inhabitants

The number of residents in a dwelling is critical for the estimation of energy consump-tion as it is stated in equation 3.8. The statistical value stated in (Senatsverwaltungfur Stadtentwicklung 2010) ,section 5.1.5, regarding the residential area per personis used for the estimation of the number of residents in a dwelling and consequentlyin a building. Nevertheless after taking into consideration the results shown in theprevious section (section 6.1.2) lead me to do the estimation considering the twomethods, which are presented below.

First Method

The number of inhabitants per dwelling is estimated basically by dividing its area bythe area per person (equation 6.5), finally the number of inhabitants per building isestimated by multiplying this result by the total number of dwellings of the building(equation 6.5).

NI = int

(

AD

Ap

)

(6.4)

NI,B = NI · ND (6.5)

Where:

• NI = Number of inhabitants per dwelling

• AD = Dwelling’s area

• Ap = Area per person

• ND = Building’s number of dwellings

• NI,B = Building’s number of inhabitants

Second Method

For the second method first the number of inhabitants per storey is estimated di-viding the constructed area by the area per person(equation 6.8), then the totalnumber of the building is estimated (equation 6.8), finally this value is divided bythe number of dwellings in that building (equation 6.8).

NI,S = int

(

CAS

Ap

)

(6.6)

NI,B = NI,S · NS (6.7)

NI = int

(

NI,B

ND

)

(6.8)

Where:

• CAS = Constructed Area of a storey (as it is defined in section 5.1.5

• NI,S = Storey’s number of inhabitants





6. Results

• NS = Building’s number of storeys

A comparison of the results obtained by the two methods are shown in table 6.6.

Method I Method II

Building ID NI DHA,YoC NI,B NI,S NI,B NI ∆I,D ∆I,B

BLDG 0003000000067203 3 63 189 11 99 2 1 90

BLDG 0003000000069010 4 21 84 6 42 2 2 42

BLDG 000300000006f0e2 5 24 120 9 54 3 2 66

BLDG 000300000006e689 3 30 90 11 66 3 0 24

BLDG 0003000f001d3542 2 21 42 4 28 2 0 14

BLDG 0003000f001f67be 4 21 84 7 49 3 1 35

BLDG 0003000e005b3d5a 3 12 36 5 30 3 0 6

BLDG 0003000000071392 2 20 40 7 35 2 0 5

BLDG 0003000f001d355a 2 12 24 4 24 2 0 0

BLDG 0003000f001d35a9 2 21 42 5 35 2 0 7

Table 6.6.: Number of inhabitants per buildingsource: Own table

I decided to continue with the second approach and this is because of the differencebetween the two of them on the final estimation of the number of residents of abuilding, which is relevant and influences directly on the values of the estimationof electrical energy consumption. Another way to evaluate the quality of results isto compare these results with the official statistical values for the borough of thearea of study (Charlottenburg-Wilmersdorf) that are presented in Ergebnisse desMikrozensus im Land Berlin 2011 (Amt fur Statistik Berlin-Brandenburg 2012),which are shown in table 6.7.

BoroughTotal 1 Person

Number of peopleaverage

Together 2 3 4/more1000 People

Mitte 202,7 126,9 75,9 45,1 15,5 15,3 1,65

Friedrichshain-Kreuzberg 166,7 104,2 62,5 36,8 13,0 12,7 1,64

Pankow 228,2 134,3 93,9 59,4 20,1 14,4 1,64

Charlottenburg-Wilmersdorf 194,9 114,3 80,7 52,9 14,9 12,8 1,64

Spandau 127,7 65,0 62,7 39,8 13,0 9,8 1,79

Steglitz-Zehlendorf 159,1 77,3 81,8 51,5 14,3 16,0 1,84

Tempelhof-Schoneberg 192,2 102,5 89,7 56,8 17,6 15,3 1,76

Neukolln 173,0 95,5 77,4 44,6 16,1 16,8 1,80

Treptow-Kopenick 137,6 68,0 69,5 47,0 12,7 9,9 1,75

Marzahn-Hellersdorf 134,3 59,2 75,1 48,5 16,7 9,8 1,86

Lichtenberg 151,9 74,3 77,6 55,0 13,4 9,1 1,73

Reinickendorf 127,1 57,2 69,9 43,6 12,0 14,3 1,91

Total Berlin 1995,4 1078,8 916,6 581,2 179,3 156,2 1,74

Table 6.7.: Households in the City of Berlin in 2011 by borough and household sizesource: Translated version of table “6.2 Privathaushalte im Land Berlin 2011 nach Bezirken und Haushaltsgroße” (Amt fur

Statistik Berlin-Brandenburg 2012)





6. Results

So considering the average number of inhabitants of the borough 1,64 and multi-plying this value for the number of dwellings give us an approximated value thatbuilding should have, values for the same buildings are shown in table 6.8.

Building ID DHA,YoC avg. Inh Mtd. I ∆Mtd. I Mtd. II ∆Mtd. II

BLDG 0003000000067203 63 103.32 189 -86 99 4

BLDG 0003000000069010 21 34.44 84 -50 42 -8

BLDG 000300000006f0e2 24 39.36 120 -81 54 -15

BLDG 000300000006e689 30 49.2 90 -41 66 -17

BLDG 0003000f001d3542 21 34.44 42 -8 28 6

BLDG 0003000f001f67be 21 34.44 84 -50 49 -15

BLDG 0003000e005b3d5a 12 19.68 36 -16 30 -10

BLDG 0003000000071392 20 32.8 40 -7 35 -2

BLDG 0003000f001d355a 12 19.68 24 -4 24 -4

BLDG 0003000f001d35a9 21 34.44 42 -8 35 -1

Table 6.8.: Comparison of the two methods for the estimation of the number ofinhabitants against the average number of resident per dwelling

source: Own Table

Where avg. Inh is the theoretical number of inhabitants based on the average valuefor the borough, the comparison results where obtained from table 6.6. Based on theresults presented on table 6.8 the decision of using Method II for the estimationof the number of inhabitants is strengthen.

6.2. Electrical Energy Consumption

The estimation of the electrical energy consumption was done based on the numberof dwellings per building (Section 6.1.2), number of inhabitants per dwelling (Section6.1.3) and the energy appliances (Section 5.2) that were assumed as presented oneach building. Another factor variable that was include was the life style proposed(Section 5.3), figure 6.13 shows the life style scenarios of several electrical appliances,for instance electrical lighting.

(a) (b)

Figure 6.13.: Life style example for the electrical lighting. Figure (a) shows the lifestyle for week days. Figure (b) shows the life style for weekends andbank holidays

source: Own graph





6. Results

Where:

• 100% Lamps are on the full time that users are at home.

• 50% Lamps are on half of the time that users are at home.

• 25% Lamps are on quarter of the time that users are at home.

• 15% Lamps are on a sixth part of the time that users at home.

Initially only three time ranges were proposed to be evaluated, nevertheless an ad-ditional one (15%) was include in order to evaluate what is the result of a low useof the appliances.

6.2.1. Electrical Energy Consumption per Building

This section presents the results per building. First the consumption of electricalenergy per day is estimated for a single dwelling, then the total value of the building iscomputed. The week value is calculated by assuming 5 labour days and two weekenddays. Week is the time framework that is used to estimated the final Year result bymultiplying the initial result by the number of weeks of a year (52). Figure 6.14shows an activity graph of the electrical energy consumption estimator used duringthe master thesis.

Figure 6.14.: Extract of the activity diagram for estimation of energy consumptionsource: Own graph

Table 6.9 presents the results of the energy consumption per building. It is possibleto see that the constructed area of the building (which in this scenario is the con-structed area of the ground floor times number of storeys) has a big influence onthe estimation of the electrical energy consumption for the methodology that wasused in this master thesis and this is because almost all parameters (except for theenergy appliances) are estimated based on the constructed area of the building.





6. Results

Building ID Const. Area High 100% Medium 50% Low 25% 15%

BLDG 0003000000067203 3589 454085.3996 231668.9144 120460.6718 75977.3748

BLDG 0003000000069010 1541 295191.7541 149802.1645 77107.36975 48029.45184

BLDG 000300000006e689 2331 421702.5059 214003.0922 110153.3854 68613.50262

BLDG 000300000006f0e2 2040 337362.0047 171202.4738 88122.70829 54890.8021

BLDG 0003000000071392 1324 168179.7776 85803.30164 44615.06364 28139.76844

BLDG 0003000e005b3d5a 961 168681.0024 85601.23688 44061.35414 27445.40105

BLDG 0003000f001d3542 968 176588.7665 90093.46673 46845.81683 29546.75687

BLDG 0003000f001d355a 716 168681.0024 85601.23688 44061.35414 27445.40105

BLDG 0003000f001d35a9 1096 176588.7665 90093.46673 46845.81683 29546.75687

BLDG 0003000f001f67be 1656 295191.7541 149802.1645 77107.36975 48029.45184

Table 6.9.: Building’s energy consumption results (kW/H Year) for different lifestyles

source: Own Table

Figure 6.15 presents the results that are shown in table 6.9.

Figure 6.15.: Building’s energy consumption results (kW/H Year) for different lifestyles

source: Own graph

The results are logical on the fact that the bigger the building the more electricalenergy it consumes, nevertheless table 6.15 shows us a interesting value that mustbe taken into consideration. A wrong estimation of the building parameters leadsto a wrong estimation of the energy consumption and this can be seen in the casesof buildings id BLDG 0003000e005b3d5a, BLDG 0003000f001d355a. Despite of thefact that they have different constructed areas, they do have the same value of con-sumption 168681.0024 kW/H year.

The results of estimation of the number of storeys of a building said that for buildingBLDG 0003000e005b3d5a , it has 6 storeys whether than in real life it has seven andfor the BLDG 0003000f001d355a the estimator says that it has 6 storeys whetherin real life it has 5 (table 6.2).





6. Results

6.2.2. Electrical Energy Consumption per Dwelling

As it is stated in figure 6.14 energy consumption values were also estimated perdwelling, this value can vary according to the total number of residents and dwellingsof the building. Table 6.10 shows the results of the estimation of annual electricalenergy consumption of dwelling in those buildings.

Building ID DHA,YoC High 100% Medium 50% Low 25% 15%

BLDG 0003000000067203 54 8408.988882 4290.165082 2230.753182 1406.988422

BLDG 0003000000069010 21 14056.7502 7133.436407 3671.779512 2287.116754

BLDG 000300000006e689 30 14056.7502 7133.436407 3671.779512 2287.116754

BLDG 000300000006f0e2 24 14056.7502 7133.436407 3671.779512 2287.116754

BLDG 0003000000071392 20 8408.988882 4290.165082 2230.753182 1406.988422

BLDG 0003000e005b3d5a 12 14056.7502 7133.436407 3671.779512 2287.116754

BLDG 0003000f001d3542 21 8408.988882 4290.165082 2230.753182 1406.988422

BLDG 0003000f001d355a 12 14056.7502 7133.436407 3671.779512 2287.116754

BLDG 0003000f001d35a9 21 8408.988882 4290.165082 2230.753182 1406.988422

BLDG 0003000f001f67be 21 14056.7502 7133.436407 3671.779512 2287.116754

Table 6.10.: Average dwelling’s energy consumption results (kW/H Year) for differ-ent life styles at the surveyed buildings

source: Own Table

The values of the electrical energy consumption of a dwelling are homogeneous andthis is due to its number of inhabitants. For that reason only three different valueswere obtained for the surveyed buildings and they are shown below for the 100%Range.

• 8,408.988882 kW/H: 1 Person Dwelling

• 14,056.7502 kW/H: 2 People Dwelling

• 17,366.52172 kW/H: 3 People Dwelling

Those results are logical based on the statement that an increase on the number ofresidents involves an non-linear increase in the estimation of the electrical energyconsumption (equation 3.8). Values are presented for a better visualisation in figure6.16.





6. Results

Figure 6.16.: Average dwelling’s energy consumption results (kW/H Year) for differ-ent life styles at the surveyed buildings

source: Own graph

6.2.3. Electrical Energy Consumption per Inhabitant

The results that are presented here (table 6.11 and figure 6.17), were estimated fromthe annual building energy consumption divided by its number of residents so a finalaverage value per person was estimated.

Building ID Bldg.InhMtd.II High 100% Medium 50% Low 25% 15%

BLDG 0003000000067203 99 4586.721208 2340.090045 1216.774463 767.4482303

BLDG 0003000000069010 42 7028.375098 3566.718203 1835.889756 1143.558377

BLDG 000300000006e689 66 6389.431907 3242.471094 1668.990687 1039.598525

BLDG 000300000006f0e2 54 6247.444532 3170.416181 1631.902005 1016.496335

BLDG 0003000000071392 35 4805.136504 2451.522904 1274.716104 803.9933841

BLDG 0003000e005b3d5a 30 5622.700078 2853.374563 1468.711805 914.8467016

BLDG 0003000f001d3542 28 6306.741662 3217.623812 1673.064887 1055.241317

BLDG 0003000f001d355a 24 7028.375098 3566.718203 1835.889756 1143.558377

BLDG 0003000f001d35a9 35 5045.393329 2574.099049 1338.451909 844.1930533

BLDG 0003000f001f67be 49 6024.321513 3057.187031 1573.619791 980.1928946

Table 6.11.: Average energy consumption per person results (kW/H Year) for differ-ent life styles at the surveyed buildings

source: Own Table





6. Results

Figure 6.17.: Average energy consumption per person results (kW/H Year) for dif-ferent life styles at the surveyed buildings

source: Own graph

Based on the fact that the number of residents of a building is not estimated ac-cording to the number of inhabitants per dwelling rather than by the number ofinhabitants per storey, the average value energy consumption of each building dif-fers from the ones that are shown in section 6.2.2.

As it has been seen on the results of the entire document I considered only in-teger values for the number of storeys, dwellings and residents. This is because atthe end there are no 1.64 residents in a dwelling in real life. They are either 1 in apositive scenario or 2 in a negative scenario. I decided to consider the negative sce-nario and use the ceiling function on all estimations for the building parameters. Forthat reason buildings such as BLDG 0003000000067203 has the following results,28 dwellings and 63 Inhabitants.





6. Results

6.3. Concrete Case of Study

A concrete building is shown in order to have a better visualisation of influence of thevariables in the forecasting done in this master thesis. The building is shown in figure6.18. For identification purposes, the building’s Id is BLDG 000300000006ca52.

Figure 6.18.: Concrete examplesource: Own graph using Apple Maps for iPad

It is important to state that it was possible to do a small poll to a resident of thatbuilding in order to compare the results obtained in the implementation, figure 6.19presents the scanned image of the poll filled by the resident.





6. Results

Figure 6.19.: Scanned image of the poll done to a residentsource:Own Graph

The building has the following parameters.

• Number of storeys: 5

• Number of ring bells at the entrance1: 20

1This value is assumed as the number of dwellings in the building, based on the fact that everydwelling has its on ring bell.





6. Results

• Year of construction range: 1900 - 1918

• Constructed Area: 536.76m2, average dwelling size: 107.35m2

• Average dwelling size according to the year of construction (table 6.4):95.27m2

It is relevant to mention two characteristics of this building, the first one is that theground floor of the building is used for commercial activities, based on this fact itcan be stated that the number of dwellings per storey is 5 (number of rings/numberof storeys) and the second one is that despite of the fact that the roof space is tallenough to be used for residential purposes that is no the case of this building. Theresults obtained of the building parameters are shown in table 6.12.

Storeys Dwellings Inhabitants per dwelling

SStd.H = 8; ǫ = −3 DStd.H = 64; ǫ = −44NI,Mtd. II = 3; ǫ = −1

SH,YoC = 6; ǫ = −1DH,Y oC = 48; ǫ = −28

DHA,YoC = 36; ǫ = −16

Table 6.12.: Case of study. Building parameters resultssource: Own Table

It is possible to see the dramatic difference in the number of dwellings of the building,this clearly indicates that the average flat area in Berlin does not fit the characteristicof this building. Furthermore, the average dwelling size according to the building yearof construction estimated during this master thesis has a significant difference withthe size of the flat that was polled, which values can be seen in figure 6.19.

Estimation of Energy Consumption

The estimation of the electrical energy consumption of the building is presented foreach of the energy appliances that were presented in section 3.3, all values wereestimated on a daily basis as it is stated in table 5.8. Table 6.13 presents the foodpreparation energy appliance (table), fridge’s consumption value assumed for theyear is divided by 365 so the daily value was obtained, due to the fact that it isalways on there is no variation on the estimation value.

Laundry group was declared as a number of cycles weekly so those values are di-vided by 7 to have the daily value (table 6.14). The electrical lighting (tables 6.15and 6.16), Entertainment and Technology (tables 6.17 and 6.18), Personal Computerand home offices (tables 6.19 and 6.20) and Miscellaneous groups (tables 6.21 and6.22) are divided into Business (week) days and weekend days based on the life stylemodel assumed in this master thesis (section 5.10).

The reason for not splitting the Food Preparation (table 6.13) and Laundry (sec-tion 6.14) energy appliances into business and weekend days stays on the fact thatthe first one is a basic need that people must supply daily and the latter one thetime scale used, which is a weekly consumption. Table 6.13 shows the high influencethat cooking has on the energy consumption of a dwelling, almost all the devices





6. Results

considered require a high amount of electricity for their operation, so an updateon the kitchen equipment will have as a direct consequence the reduction on theconsumption of electrical energy.

The difference between the life style model assumed in this master thesis and theconcrete example that is presented in this section is that the residents do not havea Chimney Hood, a device that in the worst scenario could have a consumption ofcirca 1,5 kW/h.

In the case of the Laundry energy appliance (table 6.14), its values will be reduceddramatically it the dwelling either not has a electrical tumble dryer or it has a gasbased one.

Time Range Time (h) Fridge Hob Oven Chimney Hood Microwave Dishwasher Total

100% 2 0.00112 18.88751 6.54767 1.25917 0.40000 5.03667 32.13214

50% 1 0.00112 9.44376 3.27384 0.62958 0.20000 2.51833 16.06663

25% 0.5 0.00112 4.72188 1.63692 0.31479 0.10000 1.25917 8.03387

15% 0.3 0.00112 2.83313 0.98215 0.18888 0.06000 0.75550 4.82077

Table 6.13.: Case of study. Food preparation energy appliance resultssource: Own Table

Time Range Time (cycles) Washing Machine Tumble Dryer Total

100% 2 0.37775 1.49901 1.8767650% 1 0.18888 0.74950 0.9383825% 0.5 0.09444 0.37475 0.4691915% 0.3 0.05666 0.22485 0.28151

Table 6.14.: Case of study. Laundry energy appliance resultssource: Own Table

The Electrical lighting (tables 6.15 and 6.16) is a sensitive energy appliance morethan any other. It is pretty much influenced by several factors like the room illu-mination, time of the day, user’s behaviour, as well as the economy because saverlamps are much expensive than the incandescent light bulbs. It is important to men-tion that for the model used here, the highest consumer are the 100W bulbs, so areduction on the use of those kind of bulbs reduces the energy consumption.

Time Range Time (h) 100w (2 Lamps) 40w (2 Lamps) 32w (2 Lamps) Saver Lamp (3 Lamps) Total

100% 7 2.93806 1.17522 0.94018 0.08814 5.1416

50% 3.5 1.46903 0.58761 0.47009 0.04407 2.5708

25% 1.75 0.73451 0.29381 0.23504 0.02204 1.2854

15% 1.05 0.44071 0.17628 0.14103 0.01322 0.77124

Table 6.15.: Case of study. Electrical lighting energy appliance results. Business daysource: Own Table





6. Results

Time Range Time (h) 100w (2 Lamps) 40w (2 Lamps) 32w (2 Lamps) Saver Lamp (3 Lamps) Total

100% 9 3.7775 1.511 1.2088 0.11333 6.61063

50% 4.5 1.88875 0.7555 0.6044 0.05666 3.30531

25% 2.25 0.94438 0.37775 0.3022 0.02833 1.65266

15% 1.35 0.56663 0.22665 0.18132 0.017 0.99159

Table 6.16.: Case of study. Electrical lighting energy appliance results. Weekend daysource: Own Table

The Entertainment and Technology energy appliance group (tables 6.17 and 6.18)has a special characteristic, the energy consumption of the TV (in this case a brandnew Phillips model) is lower than the energy consumption of the Home Theatre,which in this case is a home theatre system from the same brand.

The idea of using energy consumption values of the latest models was done be-cause that equipment require less energy. Notwithstanding not all the times that aTV is used involves the use of the other equipment, also the energy consumption ofthe TV includes as well the consumption of a digital TV adapter, which might notbe required by new models.

Time Range Time (h) TV on TV standby Home Theatre on Home Theatre standby Total

100% 7 2.73239 0.17170 1.54248 0.00745 4.45402

50% 3.5 1.36620 0.20705 0.77124 0.01332 2.35781

25% 1.75 0.68310 0.22473 0.38562 0.01626 1.30971

15% 1.05 0.40986 0.22014 0.23137 0.01744 0.87881

Table 6.17.: Case of study. Entertainment and Technology energy appliance results.Business day

source: Own Table

Time Range Time (h) TV on TV standby Home Theatre on Home Theatre standby Total

100% 9 3.51308 0.15150 1.98319 0.00409 5.65186

50% 4.5 1.75654 0.19695 0.99159 0.01164 2.95673

25% 2.25 0.87827 0.21968 0.49580 0.01542 1.60916

15% 1.35 0.52696 0.21379 0.29748 0.01693 1.05516

Table 6.18.: Case of study. Entertainment and Technology energy appliance results.Weekend day

source: Own Table

The results obtained in the Personal Computer and Home Office (tables 6.19 and6.20) are for a single PC per dwelling. Nevertheless this is not the scenario in thepolled dwelling, where each of the residents do have a computer. This means thatthe energy consumption of that device could be double or higher than the value thatis presented here.

The reader can see that computers have a high demand at least on the polled housebecause it is used 4h a day which in terms of the time range proposed for this masterthesis means half of the time that a person is at home. It is not clear if the valuewrote in this field express the hours used per person or the total time.





6. Results

Time Range Time (h) PC Multifunctional Modem Total

100% 7 0.88142 0.02099 0.072 0.974450% 3.5 0.44071 0.01049 0.072 0.523225% 1.75 0.22035 0.00525 0.072 0.297615% 1.05 0.13221 0.00315 0.072 0.20736

Table 6.19.: Case of study. Personal Computer and Home Office energy applianceresults. Business day

source: Own Table

Time Range Time (h) PC Multifunctional Modem Total

100% 9 1.13325 0.02099 0.072 1.2262450% 4.5 0.56663 0.01049 0.072 0.6491225% 2.25 0.28331 0.00525 0.072 0.3605615% 1.35 0.16999 0.00315 0.072 0.24514

Table 6.20.: Case of study. Personal Computer and Home Office energy applianceresults. Weekend day

source: Own Table

The final energy appliance group is called Miscellaneous (tables 6.21 and 6.22). Itpresents those devices that were not classified into the previous categories, mobilephones and tablets are part of this group and their consumption is estimated to thetime they are charged. Based on this statement, the consumption was considered asconstant as there is the tendency of users of leave their devices to charge during thenight (while users are sleeping), circa 8h a day.

The time of use of the other devices varies, for example the maximum range forvacuum and coffee machine was assume as 1h daily, 1h weekly. The iron machinetime ranges are the same as the ones in the previous energy appliances groups.

Range Time Mobile Tablet Vacuum Iron Machine Coffee Machine Docking station Total

100% 0.12 0.20147 0.38974 0.29980 1.25917 0.10283 2.37301

50% 0.12 0.20147 0.19487 0.14990 0.62958 0.05142 1.34724

25% 0.12 0.20147 0.09744 0.07495 0.31479 0.02571 0.83435

15% 0.12 0.20147 0.05846 0.04497 0.18888 0.01542 0.62920

Table 6.21.: Case of study. Miscellaneous energy appliance results. Business daysource: Own Table

Range Time Mobile Tablet Vacuum Iron Machine Coffee Machine Docking station Total

100% 0.12 0.20147 0.38974 0.29980 1.25917 0.13221 2.40239

50% 0.12 0.20147 0.19487 0.14990 0.62958 0.06611 1.36193

25% 0.12 0.20147 0.09744 0.07495 0.31479 0.03305 0.84170

15% 0.12 0.20147 0.05846 0.04497 0.18888 0.01983 0.63361

Table 6.22.: Case of study. Miscellaneous energy appliance results. Weekend daysource: Own Table

The final result is computed by the sum of the partial energy consumptions as itis stated in equation 3.19. Table 6.23 shows the final results of electrical energyconsumption in daily, weekly and monthly basis for a mentioned Dwelling. Theannual values are presented in table 6.24





6. Results

Range Time Business day Weekend day Weekly Monthly

100% 46.95192919 49.90000641 334.5596588 1338.238635

50% 23.80405689 25.2780955 169.5764754 678.3059018

25% 12.23012074 12.96714004 87.08488378 348.3395351

15% 7.588895495 8.027778283 54.00003404 216.0001362

Table 6.23.: Case of study. Electrical energy consumption of a dwelling specified byday, week and month

source: Own Table

Range Time Year Building consumption Consumption per Person

100% 17397.10226 626295.6812 626295.6812

50% 8817.976723 317447.162 317447.162

25% 4528.413957 163022.9024 163022.9024

15% 2808.00177 101088.0637 101088.0637

Table 6.24.: Case of study. Annual values of electrical energy consumption perdwelling, building and person

source: Own Table

The initial idea of having different time ranges was to evaluate the results of theworst scenario possible, which is the case that users will demand all the energyappliances at the same time when they are at home (100% or High time range).Obviously that kind of scenario is not realistic but nevertheless it gives us a goodoverview of which are the energy appliances that demand most of the energy that isconsumed in a dwelling.

At the end and based on the results obtained the 15% range time obtained theclosest results of the estimator for the polled dwelling 2780 kW/h per year againstthe 2808.00177 obtained.





7. Summary and future work


A proposal for the estimation of the electrical energy consumption of a buildingusing a 3D model is done in this master thesis, nevertheless there are certain topicsthat do need to be discussed.

A LoD2 CityGML model could be a sufficient tool for the extraction of the buildingparameters that were mentioned in this master thesis and that are required (numberof dwellings and their number of inhabitants). Nevertheless there was a lack of datain the dataset that was available that clearly influenced the hole process.

The CityGML standard indicates that some of the data that was required couldbe available by only reading the feature and its attributes, yearOfConstruction,storeysAboveGround, storeysBelowGround among other attributes were not avail-able on the dataset. That data was estimated but those results were not perfect andhad errors that should not be ignore. In certain buildings the difference betweenobserved and estimated number of storeys was of more than 2. Another way to over-come those problems could be the use of higher detailed model (LoD3 or LoD4), thelatter one is a very demanding and expensive model (time and effort) even more fora complete city or region.

An observation that can be done to the dataset used (CityMGL version 1.0) isregarding roofs. This is because some of them might have additional constructionsbut they should not connected to the building itself but rather to be part of the roofas a hole like the entrances that were shown in figure 6.5.

The CityGML model itself is not explicit neither the number of dwellings or thenumber of residents of that building, so if this kind of data is going to be stored, itshould be done as a generic attribute for both cases. Those attributes could storethe total number of inhabitants of the building as well as a number that express thetotal of dwellings of the building.

I consider that dwellings do not have an specific attribute inside the CityGML stan-dard because only the LoD4 model goes indoor in the modelling. Also the attributeterm is only a semantical concept that indicates that several rooms can be groupedby a common factor, which is that their usage is residential all of them belong tothe same residential unit.

In my opinion, the use of an average flat size and living area per person of a cityare not a good indicator of the possible number of inhabitants in a dwelling as wellthe the number of dwellings in a building. A first reason is because constructionlegislation changes between certain periods of time (decades for example). Anotherone is more related to economy factors, wealthy people can pay more for housing






meaning that their residential places are normally bigger than the average ones. Thisis the case of the area of study of this master thesis, despite of the fact that I usethe official values stated by the Senate of Berlin, they clearly do not represented theborough of Charlottenburg-Wilmersdorf.

Regarding the electrical energy consumption estimation, I consider that the End-Use method was suitable for the scope of the master thesis, nevertheless the methodshould be improved by involving a bigger and detailed database of energy appliances,despite of the fact this is could be a tedious task for instance there a hundreds ofPersonal Computers, every single of them demanding a different amount of electricalenergy for their operation.

Future Research

In future researches more detailed models should be used for the estimation of elec-trical energy consumption. Having a LoD3 model more possibilities to estimate forexample the number of storeys of a building , i.e. evaluating the number of rows ofwindows in the facade, to mention one. Also having that kind of opening (window)will allow further evaluations, for instance room illuminance based on sun lightingand how does affect the usage of electrical lighting.

As a suggestion the CityGML standard could be modify by adding specific attributesto store more information regarding residential buildings. I consider that it is rele-vant for a residential building to have the number of dwellings that contains. Alsofor those buildings that its function is mixed, it is relevant to know which parts orstoreys are dedicated to one activity and which others are used for only residentialpurposes. Several buildings of the area of study have this characteristic but it wasnot possible from the dataset to distinguish between usages.

Another variable that can be improved in future researches is the life style mod-els, discriminating people by gentle, age, occupation, or behaviour to mention somepossibilities. Only one dwelling was polled but it showed the need of a better model.One indicator of this situation was the number of bulbs at that place, in my proposalthe number was 9 but in that flat it was 26 and all of them are saver lamps. A bigsurvey should take place, so real values can be obtained and model as a consequenceis better evaluated.

Finally, it would be a interesting to develop an application domain extension (ADE)for energy aspects of a building or the topics that were dealt during this master the-sis. Having a XML schema to which a CityGML instance could be verified will leadto a better quality of the stored data as well as facilitate the exchange of them.





Bibliography

Bibliography

Amt fur Statistik Berlin-Brandenburg (2011). Fortschreibung des Wohngebaudeund Wohnungsbestandes in Berlin am 31. Dezember 2010, http:

// www. statistik-berlin-brandenburg. de/ Publikationen/ Stat_

Berichte/ 2011/ SB_ F1-1_ j01-10_ BE. pdf . Last checked, 2013-01-04.

Amt fur Statistik Berlin-Brandenburg (2012). Statistischer bericht.ergebnisse des mikrozensus im land berlin 2011, http: // www.

statistik-berlin-brandenburg. de/ Publikationen/ Stat_ Berichte/

2012/ SB_ A01-11-00_ 2011j01_ BE. pdf . Haushalte, Familien und Lebens-formen.

Architekten- und Ingenieur-Verein zu Berlin (1974). Berlin und seine Bauten, Vol.IV Teilen A, B, C of Berlin und seine Bauten, Eigenthum des Vereins.

Astrid, R., Udo, D., Aris, T. & Sabine, B. (2010). Balancing buildings and occu-pants - a holistic approach to thermal comfort and greenhouse gas emissions inmixed mode offices, Adapting to Change: New Thinking on Comfort. WindsorConference (6th : 2010 : Windsor Great Park, England), Windsor Great Park,England.

Bayerische Staatregierung (2007). Bayerische bauordnung, http: //

www. gesetze-bayern. de/ jportal/ portal/ page/ bsbayprod. psml?

showdoccase= 1&doc. id= jlr-BauOBY2007rahmen&doc. part= X . Lastchecked, 2013-30-04.

Bhattacharyya, S. C. & Timilsina, G. R. (2009). Energy demand models for policyformulation : a comparative study of energy demand models, http: // ideas.

repec. org/ p/ wbk/ wbrwps/ 4866. html (4866).

Bluejay, M. (2010). How much electricity does my television use? Last checked2013-06-11.URL: http: // michaelbluejay. com/ electricity/ tv. html# calculator

Branco, G., Lachal, B., Gallinelli, P. & Weber, W. (2004). Predicted versus observedheat consumption of a low energy multifamily complex in switzerland based onlong-term experimental data, Energy and Buildings 36(6): 543 – 555. ISSN=0378-7788.

Carbon Footprint Ltd. (n.d.). Household Energy Consumption, http: // www.

carbonfootprint. com/ energyconsumption. html . Last checked, 2013-03-05.

Carrion, D. (2010). Estimation of the energetic rehabilitation state of buildings forthe city of berlin using a 3d city model represented in citygml, Master thesis,Technische Universitat Berlin.



http://www.statistik-berlin-brandenburg.de/Publikationen/Stat_Berichte/2011/SB_F1-1_j01-10_BE.pdf



http://www.statistik-berlin-brandenburg.de/Publikationen/Stat_Berichte/2012/SB_A01-11-00_2011j01_BE.pdf



http://www.gesetze-bayern.de/jportal/portal/page/bsbayprod.psml?showdoccase=1&doc.id=jlr-BauOBY2007rahmen&doc.part=X



http://ideas.repec.org/p/wbk/wbrwps/4866.html

http://ideas.repec.org/p/wbk/wbrwps/4866.html

http://michaelbluejay.com/electricity/tv.html#calculator

http://www.carbonfootprint.com/energyconsumption.html

http://www.carbonfootprint.com/energyconsumption.html



Bibliography

Clevenger, C. M. & Haymaker, J. (2006). The impact of the building occupant onenergy modeling simulations.

Coppock, J. T. & Rhind, D. W. (1991). The History of GIS, in D. J. Maguire, M. F.Goodchild & D. W. Rhind (eds), Geographical Information Systems : Principlesand Applications, Vol. 1, Ottawa, Canada.

Cryer, J. & Chan, K. (2008). Time series analysis with applications in R, SpringerTexts in Statistics Series, Springer-Verlag New York. ISBN:9780387759593.

Cullen, K. A. (1999). Forecasting electricity demand using regression and MonteCarlo simulation under conditions of insufficient data, Master’s thesis, Collegeof Agriculture, Forestry, and Consumer Sciences, Morgantown, West Virginia.USA.

de la Beaujardiere, J. (2006). OpenGIS Web Map Server Implementation Speci-fication, Open Geospatial Consortium. Category: OpenGIS ImplementationSpecification, OGC Document: OGC 06-042.

DESTATIS (2009). Einkommen, konsum, lebensbedingungen, https:

// www. destatis. de/ DE/ ZahlenFakten/ GesellschaftStaat/

EinkommenKonsumLebensbedingungen/ Wohnen/ Tabellen/ HuG_

Wohnflaeche_ EVS. html . Statistisches Bundesamt. Last checked, 2013-01-04.

Eicker, U., Nouvel, R., Schulte, C., Schumacher, J. & Coors, V. (2012). 3d-stadtmodelle fur die warmebedarfberechnung, Fourth German-Austrian IBPSAConference, BauSIM 2012, Berlin, Germany.

FAO. (1991). Energy for Sustainable Rural Development Projects- Vol.1 a Reader,FAO (Training Materials for Agricultural Planning), Rome, Italy.URL: http: // www. fao. org/ docrep/ u2246e/ u2246e00. htm# Contents

Greenwood, P. & Nikulin, M. (1996). A Guide to Chi-Squared Testing, A Wiley-Interscience publication, Wiley.URL: http: // books. google. de/ books? id= bc8zfQSKOwIC

Haldi, F. & Robinson, D. (2010). The impact of occupants’ behaviour on urbanenergy demand, BauSim 2010, TU Vienna, Vienna, Austria.

Hamburg Senat (2005). Hamburgische Bauordnung, http: // www. hamburg. de/

contentblob/ 150654/ data/ hamburgische-bauordnung-hbauo) .pdf . Lastchecked, 2013-30-04.

Ho, E. (2009). Edward Ho’s Gadget Power Usage Page. Last checked, 2013-03-05.URL: http: // www. edho. com/ power/

IGGS (2009a). 3D Geo Database for CityGML, Institute for Geodesy & Geoinfor-mation Science, Technische Universitat Berlin. Edition 2.0.1.

IGGS (2009b). citygml4j. Last checked, 2012-13-08.URL: http: // opportunity. bv. tu-berlin. de/ software/ projects/

citygml4j/



https://www.destatis.de/DE/ZahlenFakten/GesellschaftStaat/EinkommenKonsumLebensbedingungen/Wohnen/Tabellen/HuG_Wohnflaeche_EVS.html




http://www.fao.org/docrep/u2246e/u2246e00.htm#Contents

http://books.google.de/books?id=bc8zfQSKOwIC

http://www.hamburg.de/contentblob/150654/data/hamburgische-bauordnung-hbauo).pdf

http://www.hamburg.de/contentblob/150654/data/hamburgische-bauordnung-hbauo).pdf

http://www.edho.com/power/

http://opportunity.bv.tu-berlin.de/software/projects/citygml4j/

http://opportunity.bv.tu-berlin.de/software/projects/citygml4j/



Bibliography

IGGS (2011). The citygml database 3d city db. Last checked, 2012-12-05.URL: http: // www. 3dcitybd. net

IGGS (2012). 3D City Database for CityGML, Institute for Geodesy & Geoinforma-tion Science, Technische Universitat Berlin. Addendum to the 3D City DatabaseDocumentation Version 2.0.1.

IGGS (2013). Determining energy relevant information using the semantic 3d citymodel of berlin for an energy atlas. GIS Project Winter Term 2012 / 2013.Unpublished document.

KEMA, Inc. (2010). California Residential Appliance Saturation Study, Vol. 1:Methodology, California Energy Commission. CEC-200-2010-004.

KEMA-XENERGY (2004). California Statewide Residential Appliance SaturationStudy. Final Report, California Energy Commission. Contract No. 400-04-009.

KIT (2012). Fzkviewer. Karlsruhe Institute of Technology. Institute for AppliedComputer Science. Last checked, 2012-26-04.URL: http: // www. citygmlwiki. org/ index. php/ Freeware

Kolbe, T. H., Groger, G., Nagel, C. & Hafele, K.-H. (2012). OGC City Geogra-phy Markup Language (CityGML) Encoding Standard, Open Geospatial Con-sortium. Category: OpenGIS Implementation Specification, OGC Document:OGC 12-019.

Lee, J. (2004). 3D GIS for Geo-coding Human Activity in Micro-scale Urban Envi-ronments, in M. Egenhofer, C. Freksa & H. Miller (eds), Geographic Informa-tion Science, Vol. 3234 of Lecture Notes in Computer Science, Springer BerlinHeidelberg, pp. 162–178. ISBN:978-3-540-23558-3.

Lee, Y. S., Yi, Y. K. & Malkawi, A. (2011). Simulating human behaviour and itsimpact on energy uses, Proceedings of Building Simulation, Sydney, Australia,pp. 1049–1056.

Loga, T., Diefenbach, D. N., Knissel, D. J. & Born, R. (2005). Entwicklungeines vereinfachten, statistisch abgesicherten verfahrens zur erhebung vongebaudedaten fur die erstellung des energieprofils von gebauden, Technical re-port. ISBN-Nr.: 3-932074-76-9.URL: http: // www. iwu. de/ fileadmin/ user_ upload/ dateien/ energie/

werkzeuge/ iwu-kurzverfahren_ energieprofil-endbericht. pdf

Mehra, M. & Bharadwaj, A. (2000). Demand forecasting for electricity, in T. E. R.Institute (ed.), Proceedings of the National Conference on Regulation in in-frastructure Services: progress and way forward, The Energy and ResourcesInstitute, New Delhi.

Meier, A., Rainer, L. & Greenberg, S. (1992). Miscellaneous electrical energy use inhomes, Energy Vol. 17(5): 509–518. ISSN-0360-5442.

Min.EnergiaMinas (n.d.). Guıa del Consumo y Facturacion de Energıa Electrica,http: // intranet. minem. gob. pe/ AppWeb/ DGE/ CalculoConsumo . Minis-terio de Energıa y Minas del Peru. Last checked, 2013-03-05.



http://www.3dcitybd.net

http://www.citygmlwiki.org/index.php/Freeware

http://www.iwu.de/fileadmin/user_upload/dateien/energie/werkzeuge/iwu-kurzverfahren_energieprofil-endbericht.pdf

http://www.iwu.de/fileadmin/user_upload/dateien/energie/werkzeuge/iwu-kurzverfahren_energieprofil-endbericht.pdf

http://intranet.minem.gob.pe/AppWeb/DGE/CalculoConsumo



Bibliography

Mitchell, B., Ross, J., Park, R. & Corporation, R. (1986). A Short Guideto Electric Utility Load Forecasting, Rand Report, Rand Corporation.ISBN:9780833007148.

Nagel, C., Becker, T., Kaden, R., Li, K.-J., Lee, J. & Kolbe, T. H. (2010). Re-quirements and Space-Event Modelling for Indoor Navigation, Open GeospatialConsortium. Category: OpenGIS Discussion Paper, OGC Document: OGC10-191r1.

Neufert, E., Brockhaus, M., Kister, J., Lohmann, M. & Merkel, P. (2005). Bauen-twurfslehre: Grundlagen, Normen, Vorschriften uber Anlage, Bau, Gestaltung,Raumbedarf, Raumbeziehungen, Maße fur Gebaude, Raume, Einrichtungen,Gerate mit dem Menschen als Maß und Ziel. Handbuch fur den Baufachmann,Bauherrn, Lehrenden und Lernenden, 38 edn, Vieweg+Teubner Verlag. ASIN:B002CB4VPK.

Organisation Intergouvernementale de la Convention du Metre (2006). Theinternational system of units (si). 8th edition.URL: http: // www. bipm. org/ utils/ common/ pdf/ si_ brochure_ 8_ en.

pdf

Page, J., Robinson, D. & Scartezzini, J.-L. (2007). Stochastic Simulation of OccupantPresence and Behaviour in Buildings, Building Simulation 2007, Beijing.

Philips (2013). Philips - Consumer Products. Last checked, 2013-14-05.URL: http: // www. philips. co. uk/ c/

Princenton University (2012). WordNet A lexical database for English.URL: http: // wordnetweb. princeton. edu/ perl/ webwn? s= electrical%

20energy

Recht NRW (2000). Bauordnung fur das Land Nordrhein-Westfalen,https: // recht. nrw. de/ lmi/ owa/ pl_ text_ anzeigen? v_ id=

5820031106092333838 . Last checked, 2013-20-04.

Reinhart, C. F. (2004). Lightswitch-2002: a model for manual and automated controlof electric lighting and blinds, Solar Energy 77(1): 15 – 28. issn: 0038-092X.

Safe Software Inc. (2012). Safe software. Last checked, 2012-13-08.URL: http: // www. safe. com

Santin, O. G., Itard, L. & Visscher, H. (2009). The effect of occupancy and buildingcharacteristics on energy use for space and water heating in dutch residentialstock, Energy and Buildings 41(11): 1223 – 1232. ISSN: 0378-7788.

Senatsverwaltung fur Stadtentwicklung (2005). Entwurf objektschluusselkatalogberlin (oska bln)fur die automatisierte liegenschaftskarte -alk-, http:

// www. stadtentwicklung. berlin. de/ geoinformation/ lika-auskunft/

download/ OSKABerlin. pdf . Abteilung III - Geoinformation, Vermessungund Wertermittlung.



http://www.bipm.org/utils/common/pdf/si_brochure_8_en.pdf

http://www.bipm.org/utils/common/pdf/si_brochure_8_en.pdf

http://www.philips.co.uk/c/

http://wordnetweb.princeton.edu/perl/webwn?s=electrical%20energy

http://wordnetweb.princeton.edu/perl/webwn?s=electrical%20energy

https://recht.nrw.de/lmi/owa/pl_text_anzeigen?v_id=5820031106092333838

https://recht.nrw.de/lmi/owa/pl_text_anzeigen?v_id=5820031106092333838

http://www.safe.com

http://www.stadtentwicklung.berlin.de/geoinformation/lika-auskunft/download/OSKABerlin.pdf





Bibliography

Senatsverwaltung fur Stadtentwicklung (2010). Marktubersicht, http:

// www. stadtentwicklung. berlin. de/ wohnen/ wohnenswerte_ stadt/

de/ marktuebersicht. shtml . Wohnenswerte Stadt / Wohnatlas Berlin.Einige Zahlen zum Wohnen in Berlin. Last checked, 2013-01-04.

Senatsverwaltung fur Stadtentwicklung und Umwelt (2013). Gebaudealte 1992-1993.FIS Broker WMS. Last checked, 2013-17-05.URL: http: // fbinter. stadt-berlin. de/ fb/ index. jsp

Shuvra, M. A., Rahman, M. M., Ali, A. & Khan, S. I. (2011). Modelling and Fore-casting Demand for Electricity in Bangladesh: Econometrics Model, 2011 In-ternational Conference on Economics, Trade and Development IPEDR, Vol. 7,Singapore.

Silverman, D. (2007). Energy units and conversions. Department of Physics andAstronomy, University of California, Irvine. Last checked, 2013-04-17.URL: http: // www. physics. uci. edu/ ˜ silverma/ units. html

Sony (2013). XA900 Wireless speaker. Last checked, 2013-03-05.URL: http: // www. sony. co. uk/ product/ mfi-ipod-speaker-docks/

rdp-xa900ip/ technical-specifications# tab

Sparx Systems Pty Ltd (2013). UML 2 Activity Diagram. Last checked, 2013-12-05.URL: http: // www. sparxsystems. com/ resources/ uml2_ tutorial/

uml2_ activitydiagram. html

Striatum Energy (2010). Cuanto consumen de electricidad los aparatos electricos.Last checked, 2013-03-05.URL: http: // www. gstriatum. com/ energiasolar/ articulosenergia/

244-consumo-electricidad-aparatos-electricos. html

Tsusiat Software (2011). Jts topology suite. Last checked, 2012-22-04.URL: http: // tsusiatsoftware. net/ jts/ main. html

USA, Energy Dept. (2002). U.S. Lighting Market Characterization: Volume 1:National Lighting Inventory and Energy Consumption Estimate, Vol. Vol. 1,United States. Dept. of Energy and Office of Energy Efficiency and RenewableEnergy, Building Technologies Program and Navigant Consulting, Inc.

Vretanos Panagiotis, A. (2005). Web Feature Service Implementation Specification,Open Geospatial Consortium. Category: OpenGIS Implementation Specifica-tion, OGC Document: OGC 04-094.

Wohnungsanwalt (2009). Wohnungen in berlin. Last checked, 2013-01-04.URL: http: // www. wohnungsanwalt. de/ Wohnungsanwalt/

Wohnungswirtschaft/ Statistik/ Statistik/ Wohnungen% 20Berlin. html

Wout, P., Dirk, S. & Hugo, H. (2010). Implementing realistic occupant behaviourin building energy simulations - the effect on the results of an optimisation ofoffice buildings, Proceedings of the 10th REHVA World Congress ”SustainableEnergy use in Buildings” pages:1-8, Antalya, Turkey.



http://www.stadtentwicklung.berlin.de/wohnen/wohnenswerte_stadt/de/marktuebersicht.shtml



http://fbinter.stadt-berlin.de/fb/index.jsp

http://www.physics.uci.edu/~silverma/units.html

http://www.sony.co.uk/product/mfi-ipod-speaker-docks/rdp-xa900ip/technical-specifications#tab

http://www.sony.co.uk/product/mfi-ipod-speaker-docks/rdp-xa900ip/technical-specifications#tab

http://www.sparxsystems.com/resources/uml2_tutorial/uml2_activitydiagram.html

http://www.sparxsystems.com/resources/uml2_tutorial/uml2_activitydiagram.html

http://www.gstriatum.com/energiasolar/articulosenergia/244-consumo-electricidad-aparatos-electricos.html

http://www.gstriatum.com/energiasolar/articulosenergia/244-consumo-electricidad-aparatos-electricos.html

http://tsusiatsoftware.net/jts/main.html

http://www.wohnungsanwalt.de/Wohnungsanwalt/Wohnungswirtschaft/ Statistik/Statistik/Wohnungen%20Berlin.html

http://www.wohnungsanwalt.de/Wohnungsanwalt/Wohnungswirtschaft/ Statistik/Statistik/Wohnungen%20Berlin.html



Bibliography

Yamaguchi, Y., Fujimoto, T. & Shimoda, Y. (2011). Occupant behaviour model forhouseholds to estimate high-temporal resolution residential electricity demandprofile, Proceedings of Building Simulation, Sydney, Australia, pp. 1548–1555.





A. Appendix A

A. Appendix A

The information that is contained in the DVD and memory flash that are attachedto this master thesis are

• MasterThesisCamiloLeon.zip - Contains the Eclipse Java project devel-oped in this master thesis, including all necessary files and libraries needed.

• EnergyDemandEstimator.java - This class was declared to initialise andcall all classes used in the master thesis.

• EDEBuilding.java - In this class the estimation of the building parametersis done.

• WriteFile.java - This class was declared to write the final output file whichis a txt file.

• citygml shape2citygml.fmw - This FME workbench (version 2013) wasused to integrate the year of construction for the buildings of the are of studywith the CityGML file used in this master thesis.

• MasterThesisCamiloLeon.jar - Jar file of the Eclipse Java Project

• CityGML Files - This Folder contains the original CityGML file and theresulting one after the FME process. It includes the CityGML files.

• ArcGIS - This Folder contains the ArcMap (version 10.1) file used for visuali-sation purposes, contains a shapefile with the GroundSurface of the ResidentialBuilding of the area and the results obtained in the Java Project.

The Java project requires Java 6 to work, the few indications that the user musttake into consideration is presented below.

/* ********************************************************* If you are using a CityGML file that does not have data regarding the year of construction,please comment between code lines 123-126 and uncomment code line 127 this is to avoid an errorfor not having that CityGML attribute in the file.** ********************************************************** If you are using a file that declares its SRS following the standard OCG, i.e. ¡gml:EnvelopesrsName=”urn:ogc:def:crs,crs:EPSG:X.XX:XXXX,crs:EPSG:X.XX:XXXX” srsDimension=”X”¿.* * please uncomment lines 135, 139-143, 165-166. And comment code line 167** If you are using a file that does not declare the SRS following the standard OGC please put theEPSG code manually in line 167** * *****************************************************

The previous box appears in the java class EDEBuilding.java, despite of having Java


Beneficiario de Colfuturo 2010 i



A. Appendix A

6 the user must Run or compile the project with the parameters that are presentedin figure A.1.

Figure A.1.: Run Configuration of Eclipsesource: own graph

Where:

• 1starg. CityGML Files/NuevaVersion.gml - CityGML with building features

• 2ndarg. conf/FunctionList.txt - List of function codes that indicates that thebuilding is residential

• 3rdarg. 3.15 - Standard storey height

• 4tharg. 2.4 - Average roof height (to estimate if an attic could be residentialor not)

• 5tharg. 38.8 - Average residential area of a person

• 6tharg. 70.4 - Average dwelling size

• 7tharg. 1 - Range time considered for the estimator, i.e. 1 = 100% or 0.15 =15%

• 8tharg. Results/BuildingEnergyConsumption1.txt - File where results are stored.


Beneficiario de Colfuturo 2010 ii



B. Appendix B

B. Appendix B

B.1. Survey Form

Address:Area:Number of residents:Number of rooms:Energy consumption last year:

Appliance Units Time of Use

Freezer / Fridge

Hob (4 cooking places)

Ovens

Chimney Hood

Microwave Oven

Dishwasher

Washing Machine

Tumble Dryer

Electrical Lighting (Lamps)

Lamp 100

Lamp 40

Lamp 32

Saverlamp

TV

Home Theater

Gaming System

Computer

Printer/scanner

Modem ADSL

Mobile Phone

Tablet

Docking station

Vacuum

Iron machine

Coffee machine

Digital TV Adapter

Table B.1.: Example of the survey made to the residentsSource: Own table based on (KEMA-XENERGY 2004) and section 5.2


Beneficiario de Colfuturo 2010 iii

technische universit¨at berlin - 3d.bk.tudelft.nl

Documents